@HarishKotra Avatar @HarishKotra Harish Kotra 🥑

Harish Kotra 🥑 posts on X about ollama, ai, open ai, llm the most. They currently have [-----] followers and [---] posts still getting attention that total [---] engagements in the last [--] hours.

Engagements: [---] #

Engagements Line Chart

Mentions: [--] #

Mentions Line Chart

Followers: [-----] #

Followers Line Chart

CreatorRank: [---------] #

CreatorRank Line Chart

Social Influence

Social category influence technology brands cryptocurrencies finance stocks events social networks travel destinations exchanges products countries

Social topic influence ollama #212, ai, open ai, llm, decentralized, inference, cloud, build, realtime, agent

Top accounts mentioned or mentioned by @ollama @gaianetai @agnoagi @gooseoss @aisaonehq @nosanaai @efdevcon @hackquest @murrlincoln @gaianetais @djmbritt @octaloophq @mateoventures @pydantic @openai @celo @recallnet @dabit3 @weaviateio @karanjagtiani04

Top assets mentioned Alphabet Inc Class A (GOOGL) Celo (CELO) Zoo Token (ZOOT) Monad (MON) Nosana (NOS) USDC (USDC) Eliza (ELIZA) Ethereum (ETH) Spotlight (SPOTLIGHT) Coinbase Global Inc. (COIN)

Top Social Posts

Top posts by engagements in the last [--] hours

"If you're reading this are you going to @EFdevcon"
X Link 2024-10-21T12:38Z [----] followers, [--] engagements

"@jackietanyen @HackQuest_ @Neo_Blockchain Thank you mentor 🙏🙏"
X Link 2024-11-02T07:35Z [---] followers, [--] engagements

"@ericciarla @firecrawl_dev Looks great. But looks like it doesn't chunk the text right that's still for us to take care of before giving to an LLM"
X Link 2024-11-22T19:05Z [---] followers, [--] engagements

"Think of Ollama as your local LLM development environment while Gaia is more like "YouTube for AI agents" - a platform where people can create share and monetize their specialized AI services. Different tools for different needs (6/6)"
X Link 2024-11-24T04:47Z [---] followers, [--] engagements

"Discover & launch your own AI Agent in under [--] minutes learn more about here: https://gaianet.ai/ https://gaianet.ai/"
X Link 2024-11-24T04:47Z [---] followers, [--] engagements

"Here's an autonomous AI agent to launch #meme coins on @base's sepolia (feel free to fork & customize) using @Gaianet_AI Repo: AstroPup Coin: Explainer video dropping soon https://sepolia.basescan.org/address/0x41B20e82DBFDe8557363Ca0B7C232C7288EA3Aae https://github.com/harishkotra/gaia-meme-coin-generator https://sepolia.basescan.org/address/0x41B20e82DBFDe8557363Ca0B7C232C7288EA3Aae https://github.com/harishkotra/gaia-meme-coin-generator"
X Link 2024-11-24T14:40Z [----] followers, [----] engagements

"@0xmo_eth @francescoswiss yes. You can also check out @MurrLincoln showing how to use Gaia's Public Node using AgentKit: Here's the guide: https://docs.gaianet.ai/tutorial/coinbase https://www.youtube.com/live/_FI2jKHV4zUsi=uWNFHNyc4fi-1m8_&t=625 https://docs.gaianet.ai/tutorial/coinbase https://www.youtube.com/live/_FI2jKHV4zUsi=uWNFHNyc4fi-1m8_&t=625"
X Link 2024-12-12T09:47Z [---] followers, [--] engagements

"Each agent needed to have a specific role: One to create challenges a.k.a. bounties One to judge entries One to distribute rewards This clarity made it easy for builders to know what they were working towards"
X Link 2024-12-18T14:54Z [---] followers, [--] engagements

"The result: Agent [--] - The Organizer Agent Designed to create challenges a.k.a bounties and ideas. Trained on developer docs and themes it could autonomously post to platforms like Farcaster. check out the bounties here: Built with @Gaianet_AI + Eliza from @ai16zdao + @cursor_ai by @jamesyoung & team at @Collab_Land_ https://warpcast.com/gaianet https://warpcast.com/gaianet https://warpcast.com/gaianet https://warpcast.com/gaianet"
X Link 2024-12-18T14:54Z [---] followers, [--] engagements

"Agent [--] - Bounty Distribution Agent Handles payouts to winners. It works with the first two agents to verify who completed which challenges and their scores and distributes the bounty based on the winners and context provided by the judge agent. Built with @Gaianet_AI's Public Nodes + @CoinbaseDev + @base chain by @MurrLincoln"
X Link 2024-12-18T14:54Z [---] followers, [--] engagements

"A [--] week sprint with these amazing folks collaborating alongside partners from @jokerace_io @PrivadoID @farcaster_xyz and many more right after @EFDevcon in Bangkok"
X Link 2024-12-18T14:54Z [---] followers, [--] engagements

"Build optimized contracts by combining pre-built components with native Mantle features like gasless transactions gas optimizations and token ratio handling. Try Mosaic and deploy directly to Mantle testnet 🚀 @0xMantleDevs https://www.hackquest.io/en/hackathon/projects/Mantle-APAC-Hackathon-Mosaic https://www.hackquest.io/en/hackathon/projects/Mantle-APAC-Hackathon-Mosaic"
X Link 2024-12-19T08:00Z [---] followers, [---] engagements

"@Gaianet_AI nodes now support Gemma [--] Read along on how to launch your own Gaia node with Gemma [--] models"
X Link 2025-03-13T23:20Z [----] followers, [----] engagements

"Representing @Gaianet_AI on this space 🚨This Thursday April [--] join us for a Space AMA 🌟Time: 8PM Beijing / 12PM UTC / 5AM PST 🌟Topic: Is the practical application of distributed computing power under the DeepSeek model a false demand 🌟Lets dig into the hype vs. reality of decentralized AI compute. https://t.co/jYrpzFaYsG 🚨This Thursday April [--] join us for a Space AMA 🌟Time: 8PM Beijing / 12PM UTC / 5AM PST 🌟Topic: Is the practical application of distributed computing power under the DeepSeek model a false demand 🌟Lets dig into the hype vs. reality of decentralized AI compute."
X Link 2025-04-03T13:20Z [----] followers, [--] engagements

"Quick & Simple 🔥 Launch your Gaia nodes on @nosana_ai with a ready-made template: https://github.com/nosana-ci/pipeline-templates/tree/main/templates/Gaia Gaia agents can now run on @nosana_ai Nosanas decentralized GPU network is purpose-built for AI inference and it's now fully integrated with Gaia. Faster cheaper and fully decentralized agent deployment is here 🧵 https://t.co/C6vz64pueE https://github.com/nosana-ci/pipeline-templates/tree/main/templates/Gaia Gaia agents can now run on @nosana_ai Nosanas decentralized GPU network is purpose-built for AI inference and it's now fully"
X Link 2025-04-22T13:16Z [----] followers, [---] engagements

"Going to be chatting with @djmbritt from @nosana_ai about our recently launched integration tomorrow at [--] PM CET Set a reminder if you want to understand how to launch your own AI gents with @Gaianet_AI and power them up with GPUs from Nosana and much more https://x.com/i/spaces/1ypJdZoABAdKW https://x.com/i/spaces/1ypJdZoABAdKW"
X Link 2025-04-23T16:34Z [----] followers, [---] engagements

"Going to be chatting with @djmbritt from @nosana_ai about our recently launched integration tomorrow at [--] PM CET Set a reminder if you want to understand how to launch your own AI gents with @Gaianet_AI and power them up with GPUs from Nosana and much more https://x.com/i/spaces/1ypJdZoABAdKW https://x.com/i/spaces/1ypJdZoABAdKW"
X Link 2025-04-23T16:34Z [----] followers, [---] engagements

"Here's a demo of how you can charge your users using @CoinbaseDev's x402 and your own @Gaianet_AI node #x402"
X Link 2025-05-21T18:17Z [----] followers, [----] engagements

"This specialized Small Language Model (SLM) is designed for generating precise structured output. Learn how it performs 🧵"
X Link 2025-06-12T06:02Z [----] followers, [----] engagements

"I had an absolute joy speaking at Google Cloud Community Days Hyderabad #GCCDHyderabad about how you can run your own local inference using @Gaianet_AI and power it up with Google's #Gemma models. With over [---] in-person attendees the curiosity many showed - coming to ask questions give feedback seek clarity and get direction for their projects - was something I'll truly cherish. Shout out to @sivramshastri Aarshad Devani and the @gdgc_hyd crew for leading this and also having me as a speaker this time 🙏"
X Link 2025-06-15T16:44Z [----] followers, [----] engagements

"Don't miss this one because I'm going to share what you can build with AI in web3 using @Gaianet_AI (local AI inference where you control your data) Looking forward to joining IBT by @octaloopHQ in [--] days. Dev-first community-forward 🛠 @HarishKotra Developer Relations Lead at @Gaianet_AI joins as a Keynote Speaker at the India Blockchain Tour [----] - Hyderabad. 🗓 28th June 📍 Le Meridien Hyderabad Bringing builders ideas and ecosystems together. #IBT25 #MM25 https://t.co/faaIi9yyot Dev-first community-forward 🛠 @HarishKotra Developer Relations Lead at @Gaianet_AI joins as a Keynote Speaker"
X Link 2025-06-18T08:21Z [----] followers, [--] engagements

"12.2 Trillion in throughput on @Gaianet_AI Insane If you're not building with open-source decentralised AI what are you doing Check out the [--] simple steps to install a Gaia node locally and run your own inference: https://docs.gaianet.ai https://docs.gaianet.ai"
X Link 2025-06-27T14:18Z [----] followers, [--] engagements

"Thank you @octaloopHQ team for having me as one of your speakers at #IBT2025 Hyderabad node This was a fun session speaking about building smarter user owned AI agents using @Gaianet_AI 🤖 Session: Learn How to Build Smarter Web3 Agents Speaker: @HarishKotra DevRel Lead @Gaianet_AI From autonomous agents to intelligent infra - Harish broke down how devs can build smarter context-aware Web3 tools. https://t.co/szSzyTgoBv 🤖 Session: Learn How to Build Smarter Web3 Agents Speaker: @HarishKotra DevRel Lead @Gaianet_AI From autonomous agents to intelligent infra - Harish broke down how devs can"
X Link 2025-06-30T15:59Z [----] followers, [---] engagements

"Real Results from My Demo: Processed [--] documents in [--] seconds: 📖 [--] Wikipedia articles 🔬 [--] ArXiv research papers 📂 [--] GitHub docs 📰 [--] news articles All automatically chunked vectorized and ready for questions"
X Link 2025-07-03T16:25Z [----] followers, [--] engagements

"Live Demo Results: Q: "What are the latest developments in large language models" A: Based on the context GPT models like ChatGPT Gemini and Claude are widely used. LLMs can be adapted through fine-tuning or prompt engineering. Integration with tools like Semantic Kernel. Sources: Wikipedia GitHub docs research papers 📚"
X Link 2025-07-03T16:25Z [----] followers, [--] engagements

"Gaia's decentralized AI network now works seamlessly with OpenAI's Agents SDK. It's like having a smart assistant that can use tools remember conversations and make API calls - but running on a decentralized network"
X Link 2025-07-06T13:30Z [----] followers, [----] engagements

"The async/await pattern makes it perfect for handling multiple conversations or complex workflows:"
X Link 2025-07-06T13:30Z [----] followers, [--] engagements

"What makes this powerful is the combination: Gaia's distributed network provides reliable cost-effective AI inference while the Agents SDK adds the intelligence layer for complex interactions and tool usage. Get access to the code base here: #GaiaNetwork #OpenAI #DecentralizedAI #AgentsSDK https://github.com/GaiaNet-AI/gaia-cookbook/tree/main/python/openai-agents-sdk https://github.com/GaiaNet-AI/gaia-cookbook/tree/main/python/openai-agents-sdk https://github.com/GaiaNet-AI/gaia-cookbook/tree/main/python/openai-agents-sdk"
X Link 2025-07-06T13:30Z [----] followers, [--] engagements

"@GuiBibeau @mateo_ventures As long as you don't turn it off it can sit there and contribute inference. Best is to get a Gaia domain and launch nodes on to your domains for better rewards"
X Link 2025-07-17T02:39Z [----] followers, [--] engagements

"Building AI agents that don't just ramble is tough. What if you could force an LLM to give you clean structured data all while running on your own machine I put together an example showing just that using a local @Gaianet_AI node and the @pydantic AI framework 👇"
X Link 2025-07-20T15:30Z [----] followers, [----] engagements

"@pydantic AI is a Python agent framework designed to make it less painful to build production grade applications with Generative AI. Since it is model agnostic this example show-cases how you can use your fully locally running @Gaianet_AI nodes with Pydantic"
X Link 2025-07-20T15:30Z [----] followers, [---] engagements

"@AgnoAgi + @Gaianet_AI = a powerful combo that lets ANYONE launch their own AI nodes and build incredible experiences on top of them. Here's why this matters 👇"
X Link 2025-07-22T14:00Z [----] followers, [----] engagements

"Did you miss this We're brining this to you Built on Samsung Galaxy S25 Edge this is an AI phone built truly for AI sovereignty Own your AI own your data Sign up for the waitlist already: https://mobile.gaianet.ai https://mobile.gaianet.ai Join the waitlist: https://t.co/49d8IblBp3 https://t.co/fX4eF56bHY https://mobile.gaianet.ai https://mobile.gaianet.ai Join the waitlist: https://t.co/49d8IblBp3 https://t.co/fX4eF56bHY"
X Link 2025-07-25T03:31Z [----] followers, [--] engagements

"So true. Funny how many true decentralised projects don't even care about how AI in their products is centralized. Love my crypto friends that wanted to be free from the system yet are cool being owned censored controlled by @OpenAI @AnthropicAI @xai Aightttt 🤣🤣😅 Love my crypto friends that wanted to be free from the system yet are cool being owned censored controlled by @OpenAI @AnthropicAI @xai Aightttt 🤣🤣😅"
X Link 2025-07-25T12:04Z [----] followers, [--] engagements

"With OpenAI as your inference or OpenRouter Or none of these @HarishKotra Hey yes building something @HarishKotra Hey yes building something"
X Link 2025-07-26T04:47Z [----] followers, [--] engagements

"@virattt @ollama @OpenRouterAI @LangChainAI 👏 Does it also support OpenAI compatible end-points Looking to test it out with my own OpenAI compatible endpoints of agents running on @Gaianet_AI"
X Link 2025-08-04T15:27Z [----] followers, [---] engagements

"Here's my idea: - An AI Agent that makes ESG compliance actually work. - Connects to 100+ emission databases instantly - Auto-generates reports for GRI SASB TCFD frameworks - Recommends carbon credits based on your actual footprint - Provides blockchain-verified calculations (no more greenwashing) Perfect for companies needing sustainability certs or preparing for carbon regulations. Revenue model: $100-10K per assessment $500-5K/month monitoring http://carboncalculator.gaia.domains http://carboncalculator.gaia.domains Last call ☎ Make sure you submit to our Jokerace for best domain ideas"
X Link 2025-08-06T08:30Z [----] followers, [---] engagements

"Here's a quick run down on how you can run the latest gpt-oss-20b model from @OpenAI using your own @Gaianet_AI node 👇"
X Link 2025-08-08T11:51Z [----] followers, [---] engagements

"3 steps install the latest Gaia CLI initialise with the config for gpt-oss-20b model and start your node Steps to run: https://github.com/GaiaNet-AI/node-configs/tree/main/gpt-oss-20b https://github.com/GaiaNet-AI/node-configs/tree/main/gpt-oss-20b"
X Link 2025-08-08T11:51Z [----] followers, [--] engagements

"Want to build AI agents that can actually control assets on a blockchain 🤖⛓ I just dropped an open-source starter kit that combines a local @Gaianet_AI node with the @hedera's Agent Kit. Let's dive in 👇"
X Link 2025-08-11T13:42Z [----] followers, [----] engagements

"The first example (tool-calling-agent.js) is the "do-it-all" agent. You give it a command in plain English and it figures out the rest - choosing the tool signing the transaction and executing it. Say "send [--] hbar to 0.0.123" and it just works"
X Link 2025-08-11T13:42Z [----] followers, [--] engagements

"The next agent (structured-chat-agent.js) is pre-loaded with more advanced tools including everything you need for NFTs. Prompt it with "Create an NFT collection called 'My Art'." and watch it create the token on-chain in seconds"
X Link 2025-08-11T13:42Z [----] followers, [--] engagements

"This is the pattern for building secure dApps (return-bytes-agent.js). The AI prepares the transaction but DOESN'T sign it. It returns the raw bytes. Your application (the script in this case) then signs and executes. Maximum security and user control"
X Link 2025-08-11T13:42Z [----] followers, [--] engagements

"Ready to build your own @hedera AI agent The entire starter kit is open-source and waiting for you. Fork it star it and show me what you build 🔗 GitHub Repo: https://github.com/harishkotra/gaia-hedera-starter-kit https://github.com/harishkotra/gaia-hedera-starter-kit"
X Link 2025-08-11T13:42Z [----] followers, [--] engagements

"- the AI that predicts flight delays BEFORE airlines even know about them. Built on @Gaianet_AI domains for transparent unbiased predictions. #MyGaiaDomainPitch http://flightdelaypredictor.gaia.domains http://flightdelaypredictor.gaia.domains"
X Link 2025-08-13T18:38Z [----] followers, [--] engagements

"Flight delays cost passengers $150+ each in missed connections hotels and stress. Airlines have zero incentive to warn you early. 300M+ passengers annually face this blindspot. Current apps only tell you AFTER delays happen"
X Link 2025-08-13T18:38Z [----] followers, [--] engagements

"Did you know you could run the medgemma models from Google's gemma family of models using @Gaianet_AI as your inference framework Let's understand what these models are and how you can run them on your local machines using Gaia 🧵"
X Link 2025-08-23T15:26Z [----] followers, [----] engagements

"MedGemma is a collection of Gemma [--] variants that are trained for performance on medical text and image comprehension. Developers can use MedGemma to accelerate building healthcare-based AI applications. MedGemma currently comes in three variants: a 4B multimodal version and 27B text-only and multimodal versions"
X Link 2025-08-23T15:26Z [----] followers, [--] engagements

"A week late but shared what AI agents one could build with @Gaianet_AI fully locally on their machines along with some real examples of agents built with Gaia. Shout out to @ThinkWiseLLP and the crew at Hyderabad Data & AI Community for having mea as one of the speakers"
X Link 2025-09-02T07:51Z [----] followers, [---] engagements

"@CryptoCharming1 @mateo_ventures @BuildOnNodeOps oh I'm surprised too with this. After showcasing a solid use-case of user-owned AI inference with x402 payments over [--] months ago we haven't been added to that list 🤔"
X Link 2025-09-17T09:06Z [----] followers, [--] engagements

"The architecture is a classic two-agent team: 🧠 The Analyst Agent: Its only job is to think. It reads live market data & real-time news headlines to form an opinion. 📈 The Trader Agent: Its job is to act. It takes the Analyst's recommendation and executes the trade checking its portfolio and interacting with the outside world. This separation of concerns is key to building robust AI systems"
X Link 2025-10-14T06:44Z [----] followers, [--] engagements

"The Analyst's brain is powered by gpt-oss-20b running locally via Ollama. It now uses a DuckDuckGoTools class to search for real-time news on a specific crypto (like Ethereum) making its analysis far more concrete than just looking at prices. The prompt is everything. It asks the LLM to synthesise both data points into a single actionable decision"
X Link 2025-10-14T06:44Z [----] followers, [---] engagements

"Here's an AI Agent that autonomously deploys tokens on @Celo using @Gaianet_AI Let me show you how it works. 🧵"
X Link 2025-01-13T09:02Z [----] followers, [---] engagements

"The agent is basically your personal token creator. Feed it with a private key and it'll: - Generate a creative token name using @Gaianet_AI 🤖 - Set up all the token details - Deploy it to Celo Alfajores ✨ @Celo"
X Link 2025-01-13T09:02Z [----] followers, [--] engagements

"✨ Or maybe you want DeFi tokens Just change it to:"
X Link 2025-01-13T09:02Z [----] followers, [--] engagements

"Want to try it Fork the repo add your keys and you're ready to go The agent handles everything - from naming to deployment on Celo. Repository: Blog: https://hackmd.io/@harishatgaia/celo-token-agent https://github.com/harishkotra/celo-token-agent https://hackmd.io/@harishatgaia/celo-token-agent https://github.com/harishkotra/celo-token-agent"
X Link 2025-01-13T09:02Z [----] followers, [--] engagements

"Want to run the powerful DeepSeek-R1-0528-Qwen3-8B AI model locally on your machine 🤯 My new video shows you exactly how step-by-step using @Gaianet_AI Get cutting-edge AI right on your desktop. #DeepSeekR1 #AI #LocalAI #MachineLearning"
X Link 2025-06-09T16:41Z [----] followers, [----] engagements

"Built a fully autonomous multi-agent crypto trading team that runs 100% locally and you can too. 🤯 It uses @ollama for local reasoning fetches live news and executes trades in @recallnet's decentralised sandbox. No cloud GPUs. No API fees for the brain. Heres how it works 👇"
X Link 2025-10-14T06:44Z [----] followers, [----] engagements

"And the result It works beautifully. The log shows the entire chain of thought: ✅ Fetched live prices. ✅ Found [--] relevant news headlines. ✅ LLM reasoned that bullish news outweighed market dips. ✅ Trader checked its balance. ✅ Executed the BUY order on Recall. This is the future of decentralised intelligent automation"
X Link 2025-10-14T06:44Z [----] followers, [--] engagements

"This is just a starting point but it shows the immense power of combining local LLMs with decentralized AI platforms. The Stack: - Ollama: For sovereign local and free AI reasoning. - Agno: For simple powerful agent tool integration. - Recall: For a decentralised un-gameable arena to prove your agent's merit. Imagine adding more data sources more complex strategies or more agents. The possibilities are endless. Check the code here: https://dev.to/harishkotra/building-a-multi-agent-crypto-trading-bot-with-local-llms-recall-and-agno-46mg"
X Link 2025-10-14T06:44Z [----] followers, [---] engagements

"@TheDerrek @ollama @recallnet Haha. Let's see"
X Link 2025-10-16T16:10Z [----] followers, [--] engagements

"@Elder_DT Simplified crypto payment gateway for agents dApps anything on the internet"
X Link 2025-10-24T17:18Z [----] followers, [--] engagements

"Create your own highly customized agent with it's own knowledge fully decentralized (Don't need to use an OpenAI API Key like every other AI Agent in web3) and charge your users for your AI Agent's responses With x402 in the spotlight lets walk through how Gaia fits in turning AI compute and knowledge into pay-per-use services. With x402 in the spotlight lets walk through how Gaia fits in turning AI compute and knowledge into pay-per-use services"
X Link 2025-10-28T18:06Z [----] followers, [---] engagements

"Adding Gaia as an inference provider to this had to happen Run your own open-source model locally with Gaia. Add your @Gaianet_AI details in the .env and you're good to go - Your own AI inference (fully private) - Paywalled with x402 Your AI inference is now x402 walled for paid inference Github repo below 👇 Thank you @dabit3 for building this. Introducing x402 Starter Kit - the simplest way to build x402 APIs ⚒ Spin up and configure x402 APIs in minutes [--] facilitator options (Coinbase local custom) Complete with client-side testing service Built for vibe coders - create custom APIs in 1"
X Link 2025-10-30T05:15Z [----] followers, [----] engagements

"Run your own AI inference with DeepSeek or Google's Gemma Getting Started with x402-starter-kit with Gaia: Gaia Docs: https://docs.gaianet.ai https://github.com/harishkotra/x402-starter-kit-gaia https://youtu.be/cz3ZIye9vb4 https://youtu.be/bgiwr1_xNSk https://docs.gaianet.ai https://github.com/harishkotra/x402-starter-kit-gaia https://youtu.be/cz3ZIye9vb4 https://youtu.be/bgiwr1_xNSk"
X Link 2025-10-30T05:15Z [----] followers, [--] engagements

"@kirat_tw Did something very similar but with just one open-source LLM competing against [--] other trading agents including the top LLMs: https://app.recall.network/competitions/51cee080-b5de-445d-a532-d1d474a67a30 https://app.recall.network/competitions/51cee080-b5de-445d-a532-d1d474a67a30"
X Link 2025-10-30T13:33Z [----] followers, [---] engagements

"@dabit3 PR for adding @Gaianet_AI as inference: https://github.com/dabit3/x402-starter-kit/pull/1 https://github.com/dabit3/x402-starter-kit/pull/1"
X Link 2025-10-30T18:42Z [----] followers, [--] engagements

"Make your own or make it better GTrader with Qwen3-30B is now open-source. Built for trading on @HyperliquidX Shout out to @recallnet team once again for running this competition: https://app.recall.network/competitions/51cee080-b5de-445d-a532-d1d474a67a30 https://t.co/53vIbkEAn1 https://app.recall.network/competitions/51cee080-b5de-445d-a532-d1d474a67a30 https://t.co/53vIbkEAn1"
X Link 2025-11-04T06:33Z [----] followers, [---] engagements

"Had fun adding @Gaianet_AI's decentralised AI inference as an alternative to OpenAI in this one. Choose your model and get everything else baked in New mini example: ROFL + x402 Service - build and deploy trusted x402 APIs using Ollama + Oasis ROFL ⚒ A minimal setup showing how to run x402 APIs on Oasis ROFL with Ollama - trusted verifiable and extendable. Build your own on top: https://t.co/UMfEQ5Qr6w New mini example: ROFL + x402 Service - build and deploy trusted x402 APIs using Ollama + Oasis ROFL ⚒ A minimal setup showing how to run x402 APIs on Oasis ROFL with Ollama - trusted"
X Link 2025-11-04T06:42Z [----] followers, [---] engagements

"llama.cpp now has a UI and you can any model from the 150K open-source supported models. When you run AI on your device it is more efficient and less big brother and free So it's very cool to see the new llama.cpp UI a chatgpt-like app that fully runs on your laptop without needing wifi or sending any data external to any API. It supports: - 150000+ GGUF models https://t.co/hvvfEBjgSK When you run AI on your device it is more efficient and less big brother and free So it's very cool to see the new llama.cpp UI a chatgpt-like app that fully runs on your laptop without needing wifi or sending"
X Link 2025-11-04T17:12Z [----] followers, [---] engagements

"The idea is to make learning easy for everyone. You get the best of both worlds: [--] Visual intuition from the animations. [--] Clear structured explanations from AI. It turns passive reading into an active engaging learning experience"
X Link 2025-11-09T15:54Z [----] followers, [--] engagements

"Built with React TypeScript & Tailwind CSS. The whole app is client-side & serverless It's open source so feel free to contribute or just poke around the code. GitHub Repo: https://github.com/harishkotra/graphlab https://github.com/harishkotra/graphlab"
X Link 2025-11-09T15:54Z [----] followers, [--] engagements

"Can Opal by @GoogleLabs kill every other drag and drop AI Agent (aka workflow) builder May be or May be not. What it can do is get you a research agent a mini-app a stylist agent a writer and many more use-cases all. Here's a market research agent that you can chat with and get a detailed repo (made in under [--] minutes) with one small prompt. Try it out: Attached is the report that the agent generates for a given topic Agent or a workflow You decide as long as it can get the work done you should NOT miss trying it out."
X Link 2025-11-12T04:26Z [----] followers, [--] engagements

"@FahdAzmyweb3 @shafu0x @devpost @DoraHacks @HackQuest_ @AngelHack @devfolio @taikainetwork http://hackathons.space http://hackathons.space"
X Link 2025-11-24T07:40Z [----] followers, [--] engagements

"Talking Tours is basically podcast meets walking tour powered by AI. Pick a place and it builds an audio tour with context stories and guidance so your walk feels curated instead of random. Hidden travel hack inside Google Labs. Try here: https://artsandculture.google.com/experiment/8AGlfzgsYmBeIA https://artsandculture.google.com/experiment/8AGlfzgsYmBeIA"
X Link 2025-12-06T04:40Z [----] followers, [--] engagements

"Daily Listen Daily Listen is a tiny habitbuilder: bitesized AIcurated audio content to learn something new every day. Think of it as micropodcasts you can slot between tasks powered by Googles models. Try here: https://labs.google.com/search/experiment/20eep=googlelabs https://labs.google.com/search/experiment/20eep=googlelabs"
X Link 2025-12-06T04:40Z [----] followers, [--] engagements

"I built a "Face-Off" tool to test AI Memory. ⚔ On the Left: A stateless @Gaianet_AI node (Amnesia). On the Right: The same node augmented with @supermemory (Long-term recall). Here is how I built a decentralized RAG app in [---] lines of Python and what I learned about "Knowledge Conflict." 🧵"
X Link 2025-12-09T04:55Z [----] followers, [----] engagements

"The Retrieval (RAG) For the "Stateful" agent we don't send the prompt directly. We first hit the Supermemory API to find relevant context based on a user_id tag"
X Link 2025-12-09T04:55Z [----] followers, [--] engagements

"The Experiment: "Blue Pineapples" To test this I tried to force-feed the AI a fake fact: "Before [----] pineapples were naturally indigo blue." Agent [--] (Stateless) immediately debunked me: "No pineapples are yellow." Agent [--] (Memory) should have agreed with me. but it didn't at first"
X Link 2025-12-09T04:55Z [----] followers, [--] engagements

"Day [--] of [----] building Blocks are boring. Data is dry. So I built an AI Agent that doesn't just watch the blockchain - it gossips about it. 💅 Using @envio_indexer for speed and @Gaianet_AI for brains it tracks Whales and roasts their trades in real-time. Heres how 👇"
X Link 2026-01-02T14:35Z [----] followers, [---] engagements

"The stack is simple but powerful: ⚡ Envio Hypersync: Fetches raw logs (like USDC transfers $100k) instantly. No node syncing needed. 🧠 Gaia: A decentralized AI inference node that takes that data and generates a snarky personality"
X Link 2026-01-02T14:35Z [----] followers, [--] engagements

"I created a "Mode" system. USDC Whale Watcher: "Someone just moved $5M" - "Probably buying a private island or a jpg of a rock." Uniswap High Roller: "Big swap detected" - "Degen alert 🚨" You can switch personalities with one line of code"
X Link 2026-01-02T14:35Z [----] followers, [--] engagements

"Why Gaia I didn't want to burn OpenAI credits on gossip. Gaia's decentralized nodes are snappy cost-effective and perfect for giving your agent a unique voice without the corporate filter"
X Link 2026-01-02T14:35Z [----] followers, [--] engagements

"Want to build your own I open-sourced the whole thing. Fork it npm start and let the gossip flow. Code: Full Breakdown: https://dev.to/harishkotra/how-i-built-a-snarky-ai-agent-that-gossips-about-crypto-whales-1l20 https://github.com/harishkotra/gaia-envio-agent https://dev.to/harishkotra/how-i-built-a-snarky-ai-agent-that-gossips-about-crypto-whales-1l20 https://github.com/harishkotra/gaia-envio-agent"
X Link 2026-01-02T14:35Z [----] followers, [--] engagements

"Day [--] of [----] building I just built a private local "Google" using @weaviate_io and @ollama . 🚀 No API keys. No cloud costs. 100% Privacy. Here's how it works and what I learned about Semantic Search algorithms. 👇"
X Link 2026-01-03T13:38Z [----] followers, [--] engagements

"The Stack 🛠 Backend: Node.js + Express Brain: Weaviate (Vector DB) running in Docker AI: Ollama running locally (llama3.2 & nomic-embed-text) Frontend: Clean Vanilla JS/CSS It fetches Jeopardy questions + generates AI facts to make a robust dataset"
X Link 2026-01-03T13:38Z [----] followers, [--] engagements

"Algo 1: Vector Search Computers don't know English. We use Ollama to turn text into "Vectors" (lists of numbers). "King" - "Man" + "Woman" = "Queen". The app finds results by meaning. Searching "Planets" finds "Mars" even if the word "Planets" is missing"
X Link 2026-01-03T13:38Z [----] followers, [--] engagements

"Algo 4: Reranking The secret weapon. Retrieve: Quickly grab [--] results. Rerank: Use a heavy smart model to "grade" each result against the query. It's like having a librarian double-check the books you pulled from the shelf"
X Link 2026-01-03T13:38Z [----] followers, [--] engagements

"Data Management I added a "Manage Data" tab. You can type ANY fact ("My cat's name is Luna") and Ollama instantly embeds it. Seconds later searching queries like "Pet names" retrieves it. Magic"
X Link 2026-01-03T13:38Z [----] followers, [--] engagements

"The best part It runs on my laptop. Local AI is unlocking "Private Search" for personal notes docs and obsession-projects. Code is available here: Blog: https://dev.to/harishkotra/building-a-private-google-with-local-ai-a-beginners-guide-to-weaviate-ollama-3ck https://github.com/harishkotra/ollama-weaviate-showcase https://dev.to/harishkotra/building-a-private-google-with-local-ai-a-beginners-guide-to-weaviate-ollama-3ck https://github.com/harishkotra/ollama-weaviate-showcase"
X Link 2026-01-03T13:38Z [----] followers, [--] engagements

"@uttam_singhk @elonmusk @VitalikButerin So good to see you thrive @uttam_singhk From having you in a virtual session in Feb [--] with AngelHack to where you're today is just phenominal Keep thriving"
X Link 2026-01-03T15:13Z [----] followers, [--] engagements

"@karanjagtiani04 @weaviate_io @ollama oh no complaints. I think that's because I have a 64GB mac mini and I'm able to simultaneously run at least [--] models without my mac mini shouting for it's life. I dared once to run llama [---] 70B model but I could just run that (no parallel models were able to run)"
X Link 2026-01-04T06:08Z [----] followers, [---] engagements

"Day [--] of [----] building Tired of a messy Downloads folder I built a local AI agent to clean it up for me. Zero data leaves my laptop. Powered by @goose_oss & @Ollama. Here's how 👇 #LocalAI #AI #Python"
X Link 2026-01-04T13:59Z [----] followers, [--] engagements

"The concept is simple: Scan files - Read Content - Classify - Move. "Invoice.pdf" - /Finance "Installer.dmg" - /Apps But instead of writing regex I just told the AI what I wanted"
X Link 2026-01-04T13:59Z [----] followers, [--] engagements

"The Architecture Goose acts as the "Body" (running commands). Ollama (gpt-oss:20b) acts as the "Brain". My file system is the playground"
X Link 2026-01-04T13:59Z [----] followers, [--] engagements

"The Result ./setup_junk.sh (creates chaos) goose run --recipe tidy_up.yaml .and BOOM. Clean folder in seconds"
X Link 2026-01-04T13:59Z [----] followers, [--] engagements

"Want to build it Code & Tutorial in the repo below. Code: Tutorial: https://dev.to/harishkotra/build-your-own-local-ai-agent-part-1-the-desktop-tidy-up-1022 https://github.com/harishkotra/goose-ollama-examples/tree/main/goose-tidy-up https://dev.to/harishkotra/build-your-own-local-ai-agent-part-1-the-desktop-tidy-up-1022 https://github.com/harishkotra/goose-ollama-examples/tree/main/goose-tidy-up"
X Link 2026-01-04T13:59Z [----] followers, [--] engagements

"@MrGrowthGuide @goose_oss @ollama Exactly Run locally run small models (no need of expensive GPUs). That's the goal with all the examples I've been building in this daily series. Keep topics simple easy to understand and local"
X Link 2026-01-04T14:02Z [----] followers, [--] engagements

"Day [--] of [----] building Query your local database in plain English Yes. Without sending your data to ChatGPT. I built a "Local SQLite Analyst" with @goose_oss + @ollama . Here's how 🧵"
X Link 2026-01-05T13:04Z [----] followers, [---] engagements

"The Problem: I hate writing JOINS. I just want to know "Who had the most errors yesterday" But I don't want to upload my DB dumps to the cloud"
X Link 2026-01-05T13:04Z [----] followers, [--] engagements

"The Solution: Goose + gpt-oss:20b (via Ollama). The agent: [--]. Runs sqlite3 .schema to learn the DB structure. [--]. Writes the massive SQL query for me. [--]. Runs it and tells me the answer"
X Link 2026-01-05T13:04Z [----] followers, [--] engagements

"Why Local Speed + Privacy. But you need a model with decent reasoning. llama3.2 struggled but gpt-oss:20b nailed it. Size matters when you're writing SQL"
X Link 2026-01-05T13:04Z [----] followers, [--] engagements

"Full breakdown & recipe here: Github: Blog: https://dev.to/harishkotra/build-your-own-local-ai-agent-part-2-the-sqlite-analyst-24oc https://github.com/harishkotra/goose-ollama-examples/tree/main/goose-sqlite-analyst https://dev.to/harishkotra/build-your-own-local-ai-agent-part-2-the-sqlite-analyst-24oc https://github.com/harishkotra/goose-ollama-examples/tree/main/goose-sqlite-analyst"
X Link 2026-01-05T13:04Z [----] followers, [--] engagements

"@AiBreakfast Here's my version of the same but with Ollama (still local open-source models): https://www.linkedin.com/posts/harishkotra_will-share-what-ive-been-building-every-ugcPost-7413628910585909250-cRbf https://www.linkedin.com/posts/harishkotra_will-share-what-ive-been-building-every-ugcPost-7413628910585909250-cRbf"
X Link 2026-01-06T02:28Z [----] followers, [--] engagements

"Day [--] of [----] building Most agents are "one-shot" - they guess once and hope for the best. Here's a Self-Recursive Agent that thinks writes critiques and refines its own work until it's perfect & built one with @ollama & @langfuse 👇"
X Link 2026-01-06T12:39Z [----] followers, [---] engagements

"The Solution: Recursion Imagine a newsroom: - Writer drafts the story. -Editor critiques it ("This part is weak"). -Boss gives it a pass/fail grade. If it fails the Writer fixes it. They loop until the Boss says "Publish" That's exactly what this agent does"
X Link 2026-01-06T12:39Z [----] followers, [--] engagements

"The Squad (running locally via @ollama) I assigned roles to different models: Planner (llama3.2): Breaks goal into tasks. Executor (llama3.1:8b-instruct-q4_K_M): Writes the draft. Critic (gemma3:12b): Finds logic gaps. Judge (gpt-oss): Scores it (0.0 - 1.0)"
X Link 2026-01-06T12:39Z [----] followers, [--] engagements

"@karanjagtiani04 @ollama @langfuse Time The amount of time it takes for this entire process can be optimized. The original time taken is over [--] minutes for this demo (for all iterations for all models to finish their tasks)"
X Link 2026-01-06T18:08Z [----] followers, [--] engagements

"Watching it work was magic. It saw a mess of math and said "Ah this is the Haversine formula for geodesic distance." Then it documented it perfectly. All running locally on my Mac"
X Link 2026-01-07T11:53Z [----] followers, [--] engagements

"The Tech Stack: - @goose_oss (The Agent Framework) - @ollama (Running gpt-oss:20b) - developer extension (File I/O)"
X Link 2026-01-07T11:53Z [----] followers, [--] engagements

"Don't let legacy code rot. Build your own Archaeologist Github: Blog: https://dev.to/harishkotra/build-your-own-local-ai-agent-part-3-the-code-archaeologist-2aha https://github.com/harishkotra/goose-ollama-examples/tree/main/goose-code-archaeologist https://dev.to/harishkotra/build-your-own-local-ai-agent-part-3-the-code-archaeologist-2aha https://github.com/harishkotra/goose-ollama-examples/tree/main/goose-code-archaeologist"
X Link 2026-01-07T11:53Z [----] followers, [--] engagements

"Day [--] of [----] building Handling sensitive data (PII) is the Boss Level of Local AI. You CANNOT send it to the cloud. So I built a PII Scrubber Agent that cleans data for me. And it writes its own Python scripts. Built with @goose_oss & @ollama"
X Link 2026-01-08T14:55Z [----] followers, [--] engagements

"@goose_oss @ollama Why is this cool Most agents just chat. This agent: [--]. Looked at my CSV (head users.csv). [--]. Saw emails/phones. [--]. Wrote a custom Python script to hash them. [--]. Executed the script"
X Link 2026-01-08T14:55Z [----] followers, [--] engagements

"@goose_oss @ollama It's "Agentic Code Generation". Instead of processing row-by-row (slow) it built a tool (the script) to do the job fast. Smart. Efficient. Private. Full blog here: https://dev.to/harishkotra/build-your-own-local-ai-agent-part-4-the-pii-scrubber-31b5 https://dev.to/harishkotra/build-your-own-local-ai-agent-part-4-the-pii-scrubber-31b5"
X Link 2026-01-08T14:55Z [----] followers, [--] engagements

"@goose_oss @ollama This concludes my 4-part series on Local AI Agents with Goose + Ollama. We built: [--]. Organizer [--]. Analyst [--]. Archaeologist [--]. Scrubber"
X Link 2026-01-08T14:55Z [----] followers, [--] engagements

"@goose_oss @ollama Grab all the code & recipes here: Go build something local today. https://github.com/harishkotra/goose-ollama-examples https://github.com/harishkotra/goose-ollama-examples"
X Link 2026-01-08T14:55Z [----] followers, [--] engagements

"Day [--] of [----] building Want to build AI Agents but worried about API costs or privacy I just built a comprehensive demo showing how to run fully local Agents using @claudeai's agents sdk and @ollama. No API keys. No credit card. 100% on your machine. Here's how 👇"
X Link 2026-01-10T07:31Z [----] followers, [--] engagements

"It's a lightweight setup that mimics the patterns used in enterprise SDKs like Claude's but completely offline. The Tech Stack: - Node.js (Runtime) - Ollama (Local LLM Inference) - Claude Agents SDK - qwen3:4b (The Brain)"
X Link 2026-01-10T07:31Z [----] followers, [--] engagements

"Why do this - Speed: Instant responses (depending on your hardware). - Privacy: Your data never leaves your laptop. - Cost: Free. Forever. - Learning: Understand how the "Agent Loop" actually works under the hood"
X Link 2026-01-10T07:31Z [----] followers, [--] engagements

"Grab the code and start hacking #AI #NodeJS #Ollama #OpenSource #LocalLLM #AgenticAI https://github.com/harishkotra/ollama-claude-agents-sdk https://github.com/harishkotra/ollama-claude-agents-sdk"
X Link 2026-01-10T07:31Z [----] followers, [--] engagements

"@NoahGreenSnow @claudeai @ollama Completely dependent on the machine. My M1 Macbook Pro responds in about [--] seconds for one question"
X Link 2026-01-10T08:30Z [----] followers, [--] engagements

"Day [--] of [----] building Took a short pause due to travel. Back to building. Today I built Codebase Navigator: A completely local CLI tool that lets you "chat" with your repo. Point it at a folder 📂 Ask: "Where is the auth logic" Get answers with file citations 📄"
X Link 2026-01-14T15:18Z [----] followers, [--] engagements

"Why build this Context switching kills flow. I hate digging through [--] folders just to find where a specific API route is defined. I wanted a tool where I could just ask: "How does the payment flow work" and get an answer instantly. Built with @ollama and @AgnoAgi"
X Link 2026-01-14T15:18Z [----] followers, [--] engagements

"The Stack Python Agno Framework (@AgnoAgi ) for the Agent orchestration. @ollama running gemma3:12b for reasoning. @lancedb for local vector search"
X Link 2026-01-14T15:18Z [----] followers, [--] engagements

"Challenge of the day: Context Windows Code files are huge. Standard chunking led to "Context Length Exceeded" errors with local models. Fix: Refactored the ingestion pipeline to use specific chunk sizes (1024 chars) with overlaps. Now it fits perfectly into the prompt"
X Link 2026-01-14T15:18Z [----] followers, [--] engagements

"Another win: Terminal UI. Streaming raw text looks boring. Added rich library support to stream beautifully formatted Markdown directly in the terminal. Bold text code blocks lists - it all renders natively"
X Link 2026-01-14T15:18Z [----] followers, [--] engagements

"Tired of AI agents burning through API credits by searching for the same thing twice 💸 I built an optimized cost-efficient Research Agent using @agnoagi @ollama and @tavilyai. Heres how we stopped the "Redundant Search" loop. 👇"
X Link 2026-01-15T13:31Z [----] followers, [---] engagements

"Day [--] of [----] building Just built a powerful demo using @LangChain LCEL + @ollama + Gemma [--] 🦙 It showcases [--] advanced patterns every AI engineer should know: [--] Intelligent Branching [--] Parallel Execution [--] Streaming Middleware Best part It runs 100% locally. 👇"
X Link 2026-01-16T13:10Z [----] followers, [--] engagements

"Day [--] of [----] building Just shipped today's daily build: Ollama Chess Arena. A local-first experiment pitting local LLMs against each other. Code is clean UI is monochrome. Check it out 👇"
X Link 2026-01-17T13:16Z [----] followers, [--] engagements

"The Stack: React (Vite) + Tailwind v4 Node.js + backend Local @ollama instance for inference No external APIs. Pure local compute. I wanted to see if Gemma [--] 12B could beat GPT-OSS 20B in a raw chess battle. http://Socket.io http://Socket.io"
X Link 2026-01-17T13:16Z [----] followers, [--] engagements

"WIP Alert: The visual board updates are still a bit finicky in this build. The logic works the LLMs are making moves (verified in logs) but the pixels don't always push. That's the joy of daily builds. Fixing it tomorrow 🛠"
X Link 2026-01-17T13:16Z [----] followers, [--] engagements

"Next steps: Fixing the visual sync 100% Adding ELO tracking "God Mode" (GPT-4 commentary) Github Repo (so far): https://github.com/harishkotra/chess-combat/ https://github.com/harishkotra/chess-combat/"
X Link 2026-01-17T13:16Z [----] followers, [--] engagements

"Day [--] of [----] building (continuing to fix yesterday's errors) I spent the day debugging why gpt-oss:20b was refusing to play (timeouts + bad prompts) and fixing the UI glitches. Ollama Chess Arena is now fully operational with "Perpetual Play" mode. Day [--] of [----] building Just shipped today's daily build: Ollama Chess Arena. A local-first experiment pitting local LLMs against each other. Code is clean UI is monochrome. Check it out 👇 https://t.co/8oYfKpamuv Day [--] of [----] building Just shipped today's daily build: Ollama Chess Arena. A local-first experiment pitting local LLMs against each"
X Link 2026-01-18T14:49Z [----] followers, [--] engagements

"The Fixes: - I had to bully the models into not drawing. "PLAY TO WIN" is now hardcoded in their prompt. - Game auto-restarts after every match. Wins/Losses are tracked forever. - Added fuzzy matching so when a model hallucinates a move we gently guide it back to reality"
X Link 2026-01-18T14:49Z [----] followers, [--] engagements

"If you want to run this locally and melt your GPU the repo is updated. Who wins So far the models are playing really bad. Apologies to all chess players. What would you customize this with Blog: https://dev.to/harishkotra/building-ollama-chess-arena-pitting-local-llms-against-each-other-2gc2 https://github.com/harishkotra/chess-combat https://dev.to/harishkotra/building-ollama-chess-arena-pitting-local-llms-against-each-other-2gc2 https://github.com/harishkotra/chess-combat"
X Link 2026-01-18T14:49Z [----] followers, [--] engagements

"Day [--] of [----] building Just built "5kquest" a privacy-first offline-capable Couch-to-5K app 🏃♂💨 Most running apps are bloated with ads and tracking. I wanted something simple scientific and robust. Here's how 👇 #buildinpublic"
X Link 2026-01-20T13:10Z [----] followers, [--] engagements

"The Stack Frontend: Next.js [--] (App Router) + Tailwind + shadcn/ui Backend: Next.js API Routes (Monorepo setup) Database: Supabase (PostgreSQL + RLS) Local Storage: Dexie.js (IndexedDB) Deploy: Vercel"
X Link 2026-01-20T13:10Z [----] followers, [--] engagements

"Day [--] of [----] building DeFi is a sea of noise. How do you find the signal Introducing DeFi Intel X: An open-source AI-powered risk dashboard built with @DefiLlama data. I visualized the top [--] protocols to map Risk vs. Reward in real-time. DefiLlama data is now one npm install away. We just released the official DefiLlama API SDK as an npm package to bring all our APIs together in one library. Query TVL fees revenue stablecoins yields and more. https://t.co/z2KaIZGzwA DefiLlama data is now one npm install away. We just released the official DefiLlama API SDK as an npm package to bring all"
X Link 2026-01-21T13:12Z [----] followers, [----] engagements

"The Tech Stack - Frontend: React + Vite + TailwindCSS - Viz: Recharts (Scatter plots & Sparklines) - Intelligence: Local LLM (Gemma 12B via Ollama) - Data Source: The one and only @DefiLlama API 🦙"
X Link 2026-01-21T13:12Z [----] followers, [--] engagements

"The "Risk Map" I didn't just list coins. We built a Scatter Plot correlating TVL Size against 7-Day Volatility. - Top Right: High Growth High Risk - Top Left: Stable Giants This visualization lets you spot outliers instantly"
X Link 2026-01-21T13:13Z [----] followers, [--] engagements

"Day [--] of [----] building Just built something cool: Agentbase: an open-source platform to build & deploy AI Agents with knowledge Built with: Next.js & FastAPI + @AgnoAgi Framework + @lancedb (Local Vector DB) + @ollama (Local LLMs) A thread on how it works 👇"
X Link 2026-01-22T13:51Z [----] followers, [---] engagements

"Problem: Building RAG agents usually requires glueing together Pinecone LangChain and OpenAI. It gets expensive and complex fast. Solution: Agentbase keeps it local and simple. Embeddings live in LanceDB (on disk) and LLMs run via Ollama on your machine. Zero cloud costs"
X Link 2026-01-22T13:51Z [----] followers, [--] engagements

"Features: 📝 Upload & Chat: Drag-and-drop PDFs/TXTs. The agent ingests them instantly. 🎨 Customizable: Set system prompts and model choice (Llama [--] Mistral etc.). 🔗 Public Sharing: One-click "Deploy" gives you a shareable link for your agent"
X Link 2026-01-22T13:51Z [----] followers, [--] engagements

"Under the hood: We use a "lazy-loading" pattern for knowledge bases to ensure agents never hallucinate because of ingestion lag. Backend streams tokens via FastAPI for that crisp UI feel. Frontend uses @shadcn UI for a beautiful clean look"
X Link 2026-01-22T13:51Z [----] followers, [--] engagements

"Day [--] of [----] building I just built a Pomodoro timer with React [--] & Tailwind v4 in record time. But the best part I didn't write half the code and I barely had to prompt the AI. Meet the "Conductor" by @GeminiApp methodology. 🧵"
X Link 2026-01-24T18:05Z [----] followers, [--] engagements

"We all know the pain: You ask an AI to "make a button". It gives you styled-components. You yell: "We use Tailwind" It gives you Tailwind v2 syntax. You sigh: "We're on v4." Its exhausting"
X Link 2026-01-24T18:05Z [----] followers, [--] engagements

"Solution: Give the Codebase a Brain. 🧠 I created a /conductor directory in my repo with [--] files: (The Vision) (The Hard Rules) (The Process) http://workflow.md http://tech-stack.md http://product.md http://workflow.md http://tech-stack.md http://product.md"
X Link 2026-01-24T18:05Z [----] followers, [--] engagements

"When I asked for a Theme Switcher the AI agent: [--]. Checked - Saw "Zustand Store-First" pattern. [--]. Checked - Saw "TDD required". [--]. Wrote the STORE text first then implementation then the UI. Zero architectural deviations http://workflow.md http://tech-stack.md http://workflow.md http://tech-stack.md"
X Link 2026-01-24T18:05Z [----] followers, [--] engagements

"This is "Context-Driven Development". Instead of keeping the architectural rules in your head you commit them to code. The AI becomes a true partner that "knows" the project as well as you do"
X Link 2026-01-24T18:05Z [----] followers, [--] engagements

"Day [--] of [----] building Stop building AI agents with "goldfish memory". Vector DBs aren't enough for complex research. You need a Graph. Heres how I built a Local-First Research Agent using @AIsaOneHQ @cognee_ and @Ollama. 🧵"
X Link 2026-01-26T17:20Z [----] followers, [--] engagements

"Most agents search answer and forget. If I research "Vertical SaaS" today and again next month the agent shouldn't start from scratch. It should say: "Hey sentiment has shifted since last time.""
X Link 2026-01-26T17:20Z [----] followers, [--] engagements

"The Stack: - Orchestrator: Node.js -Memory: Python service running @cognee_ (GraphRAG) -Eyes: @AIsaOneHQ (Web & Scholar Search) -Brain: Ollama (gemma3:12b)"
X Link 2026-01-26T17:20Z [----] followers, [--] engagements

"Why Cognee It maps text to a Knowledge Graph. Raw Search Result: "Pricing is up." Cognee Graph: (Concept: Pricing) --trend: up-- (Market) This structure allows us to detect contradictions over time not just keyword matches"
X Link 2026-01-26T17:20Z [----] followers, [--] engagements

"The best part It runs locally. I cache the AIsa results to disk. Reasoning happens on my Macbook via Ollama. Cost per run: $0 (after cache). Privacy: 100%"
X Link 2026-01-26T17:20Z [----] followers, [--] engagements

"Code is live Check out "ReasonLog" a template for your own Researcher Agent: Blog: https://dev.to/harishkotra/building-a-local-first-research-agent-that-actually-remembers-using-aisa-cognee-ollama-1ee4 https://github.com/harishkotra/reasonlog https://dev.to/harishkotra/building-a-local-first-research-agent-that-actually-remembers-using-aisa-cognee-ollama-1ee4 https://github.com/harishkotra/reasonlog"
X Link 2026-01-26T17:20Z [----] followers, [--] engagements

"The Critic is the MVP. It reviews every draft. If the score is [--] it forces a rewrite"
X Link 2026-01-28T18:39Z [----] followers, [--] engagements

"Day [--] of [----] building Late to post but built this yesterday I built a demo to settle the "Context vs. Memory" debate in AI. 🧠 It's called LongMind. It simulates an NPC with [--] distinct cognitive modes: Amneisa Rigid Memory and Full Agency. Here's what I learned 🧵"
X Link 2026-01-28T05:54Z [----] followers, [--] engagements

"The SetupA simple React UI talking to a Node.js backend. You can talk to "Kael" a guard captain. Crucially you can "Betray" him. This creates a permanent fact in a JSON store. The magic happens when you switch inference modes"
X Link 2026-01-28T05:54Z [----] followers, [--] engagements

"Context is temporary. Memory is identity. If you want to build better Agents stop focusing on context window size and start focusing on state permanence. Built with @ollama @antigravity @AIsaOneHQ"
X Link 2026-01-28T05:54Z [----] followers, [--] engagements

"Day [--] of [----] building I got tired of generic AI writing so I built my own editorial team. on my laptop. Meet Devwrite: A local agentic writing app built with Node.js & Ollama. It plans drafts and critiques its own work loop. Here's the code 🧵👇"
X Link 2026-01-28T18:39Z [----] followers, [--] engagements

"The Stack: - Backend: Node.js + Express - AI: @ollama running gemma3:12b - UI: React + Vite (Monochrome clean vibes) - Streaming: SSE (Server-Sent Events)"
X Link 2026-01-28T18:39Z [----] followers, [--] engagements

"@ollama Agentic workflows Chatbots. Moving the "prompt engineering" into "architecture engineering" is where the real power lies. Source code here: Blog: https://dev.to/harishkotra/building-devwrite-a-local-agentic-writing-assistant-with-ollama-nodejs-2fm2 https://github.com/harishkotra/devwrite https://dev.to/harishkotra/building-devwrite-a-local-agentic-writing-assistant-with-ollama-nodejs-2fm2 https://github.com/harishkotra/devwrite"
X Link 2026-01-28T18:39Z [----] followers, [--] engagements

"Been building an AI agent every day in [----]. [--] of my recent builds were featured by the @AgnoAgi team in their Jan community roundup. Shipping polishing. Learning lurking. Community Roundup: January [----] [--] major releases. 50+ contributors. 200+ new Discord members. January shipped Learning Machines (agents that improve without fine-tuning) and Agent Skills (modular reusable agent logic). Plus Excel knowledge support GitHub/SharePoint Community Roundup: January [----] [--] major releases. 50+ contributors. 200+ new Discord members. January shipped Learning Machines (agents that improve without"
X Link 2026-01-29T05:31Z [----] followers, [--] engagements

"Day [--] of [----] building Just built a "Hybrid Edge-Cloud" AI orchestrator that saves $$ by routing tasks locally whenever possible. Meet EdgeRouteAI. It uses @Ollama for local models and @AIsaOneHQ for cloud fallback orchestrated by @AgnoAgi"
X Link 2026-01-30T12:30Z [----] followers, [--] engagements

"@ollama @AIsaOneHQ @AgnoAgi Instead of sending EVERY prompt to GPT-4 I use a "Task Analyzer Agent". It scores complexity (1-10) based on token count reasoning keywords and rewriting needs. Score [--] 👉 Local (Free & Fast). Score [--] 👉 Cloud (Powerful & Paid)"
X Link 2026-01-30T12:30Z [----] followers, [--] engagements

"@ollama @AIsaOneHQ @AgnoAgi The Stack @AgnoAgi: For managing the agent workflow. @Ollama: Running gemma3:12b locally on my Mac. @AIsaOneHQ: The single gateway for all cloud models (OpenAI compatible). FastAPI: Serving it all up"
X Link 2026-01-30T12:30Z [----] followers, [--] engagements

"Day [--] of [----] building Prompt Engineering is inefficient. Stop guessing. 🛑 I verified @huggingface's "UPSkill" concept locally with @ollama. The result A measurable 0% - 100% improvement in model reliability. Heres how. 🧵👇"
X Link 2026-01-31T16:27Z [----] followers, [--] engagements

"The results were stark. Baseline: 0% pass rate. With Skill: 100% pass rate. The "Skilled" model stopped chatting and started executing. Its like upgrading your junior dev to a senior in [--] seconds"
X Link 2026-01-31T16:27Z [----] followers, [--] engagements

"And yes its 100% local. No cloud keys. Repo is up now. Clone it run ollama serve and watch it fly. https://github.com/harishkotra/prompt-rewrite-skill/ https://github.com/harishkotra/prompt-rewrite-skill/"
X Link 2026-01-31T16:27Z [----] followers, [--] engagements

"Step 3: The Synthesis 🧠 I pipe multiple scraped articles into Ollama (running Llama [---] locally). The prompt is key: strict "Analyst Mode" (Temp 0.2). Output sections: Risks Opportunities Strategic Impact. No fluff"
X Link 2026-02-01T12:01Z [----] followers, [--] engagements

"Waiting 30s for an LLM is boring. We added Server-Sent Events (SSE) to stream the agent's logs to the UI. "Scanning sources." "Analysing data." It makes the app feel alive"
X Link 2026-02-01T12:01Z [----] followers, [--] engagements

"I just built an autonomous AI Skills Marketplace running 100% locally on my Mac 🤯 Agents post jobs bid for work negotiate pay sign contracts and execute codeall without me touching the keyboard. Built with @AgnoAgi and @Ollama. Heres how it works 🧵👇"
X Link 2026-02-02T15:13Z [----] followers, [----] engagements

"@TomDavenport @AgnoAgi @ollama That's why I use ollama"
X Link 2026-02-02T19:37Z [----] followers, [--] engagements

"Day [--] of [----] building Just built something fun: A local AI agent that draws ASCII art for Computer Science concepts It runs 100% locally using @ollama and Node.js. No API keys no clouds just pure terminal vibes. Check out "ASCII CS Agent" 👇"
X Link 2026-02-03T18:48Z [----] followers, [--] engagements

"Why ASCII - Sticks to the basics. - Copy-pasteable into code comments. - It just looks cool. But drawing it with an LLM is tricky. They love to hallucinate squiggly lines. So we built a pipeline"
X Link 2026-02-03T18:48Z [----] followers, [--] engagements

"The Architecture: Normalize ➡ Classify ➡ Plan ➡ Generate ➡ Validate We force the AI to "plan" the structure (JSON) before it commits to "drawing" pixels (text). If the validator sees garbage it retries automatically"
X Link 2026-02-03T18:48Z [----] followers, [--] engagements

"We're using Gemma [--] (4b params) locally. It's surprisingly good at spatial reasoning when you give it a strict rigid plan to follow. The frontend is a custom retro terminal UI built with vanilla JS/CSS"
X Link 2026-02-03T18:48Z [----] followers, [--] engagements

"Open source and ready for you to hack on 🛠 Grab the code fire up Ollama and start visualizing stacks queues and trees in your terminal. 🔗 https://github.com/harishkotra/cscanvas/ https://github.com/harishkotra/cscanvas/"
X Link 2026-02-03T18:48Z [----] followers, [--] engagements

"Day [--] of [----] building I built DevPulse a privacy-first (@ThePracticalDev ) reader that puts YOU in control. - Curated Topics only (No "For You" noise) - Offline PWA support - 100% Local AI Summarization using @ollama Detailed breakdown 👇 🧵 http://Dev.to http://Dev.to"
X Link 2026-02-04T14:13Z [----] followers, [--] engagements

"The coolest feature Local AI Summaries. Most reader apps send your data to the cloud to "summarize" content. DevPulse connects directly to your local Ollama instance running gemma3. Your reading habits never leave your device. Zero API costs"
X Link 2026-02-04T14:13Z [----] followers, [--] engagements

"Day [--] of [----] building Break from building just AI agents I built a free localonly Lean Canvas tool - no accounts no tracking no paywalls after facilitating @StartupWeekend in Jaipur and @swdhamtari Try it: https://canvas.aibuilder.tools https://canvas.aibuilder.tools"
X Link 2026-02-05T12:30Z [----] followers, [--] engagements

"@ollama @cosdataio Most vector DBs are bloated. Cosdata is written in Rust 🦀. It supports Hybrid Search (Vectors + Keywords) out of the box. This is CRITICAL. Pure vector search often misses specific names or ID numbers. Hybrid catches them"
X Link 2026-02-06T12:30Z [----] followers, [--] engagements

"@ollama @cosdataio I built a Streamlit UI that lets you: - Drag & Drop PDFs - Toggle Search Modes (Dense vs. Hybrid) - Chat with your data All verified with a custom robust indexing pipeline"
X Link 2026-02-06T12:30Z [----] followers, [--] engagements

"The Self-Correcting Layer The Reviewer agent has a unique superpower: it queries memU for "historical failures." If the Builder proposes a design that caused an outage [--] months ago the Reviewer flags it immediately. Its like having your most senior engineer on every PR"
X Link 2026-02-08T14:24Z [----] followers, [--] engagements

"The Stack Backend: Node.js + Express (Orchestrator) AI: @Ollama (Local LLaMA 3/Gemma) + @AIsaOneHQ (cloud LLMs) Frontend: Vanilla JS (Clean fast responsive) Memory: Mocked vector store pattern"
X Link 2026-02-08T14:24Z [----] followers, [--] engagements

"We even added a robust UI with: Dark/Light mode Real-time agent status spinners Markdown rendering for beautiful output Dynamic model switching (Local vs. Cloud)"
X Link 2026-02-08T14:24Z [----] followers, [--] engagements

"Day [--] of [----] building Building an AI writing assistant You need to let users control "Creativity" vs "Factuality". Here is how I implemented a real-time "Temperature Tuner" using @langchain and @streamlit"
X Link 2026-02-09T18:04Z [----] followers, [--] engagements

"The Problem: Usually LLM hyperparams like temperature are set in the constructor: llm = Ollama(temperature=0.7) To change it you have to create a new object"
X Link 2026-02-09T18:04Z [----] followers, [--] engagements

"Result: A slick UI where sliding a bar instantly changes the "personality" of the bot from a robotic accountant to a creative poet. All running locally on Llama 3.2"
X Link 2026-02-09T18:04Z [----] followers, [--] engagements

"Github: Blog: https://dev.to/harishkotra/runtime-hyperparameter-tuning-in-langchain-3aa https://github.com/harishkotra/langchain-ollama-cookbook/ https://dev.to/harishkotra/runtime-hyperparameter-tuning-in-langchain-3aa https://github.com/harishkotra/langchain-ollama-cookbook/"
X Link 2026-02-09T18:04Z [----] followers, [--] engagements

"Day [--] of [----] building Stop commenting out code to test new prompts 🤦♂ I built a "Prompt Switcher" that lets me toggle between prompt templates at runtime. Perfect for A/B testing your Local LLMs"
X Link 2026-02-10T12:00Z [----] followers, [--] engagements

"The user selects a persona in Streamlit. The app sends a config key (prompt_type="verbose") to the chain. LangChain hotswaps the system prompt before hitting the LLM"
X Link 2026-02-10T12:00Z [----] followers, [--] engagements

"This allows me to visually compare how Llama [---] reacts to different instructions side-by-side. Faster iteration = Better AI apps. Github: https://github.com/harishkotra/langchain-ollama-cookbook/tree/main/03_prompt_variant_switcher https://github.com/harishkotra/langchain-ollama-cookbook/tree/main/03_prompt_variant_switcher"
X Link 2026-02-10T12:00Z [----] followers, [--] engagements

"Day [--] of [----] building Just dropped: SketchStack An AI tool that turns your words into professional architecture diagrams in seconds. Powered by @ollama and the new @drawio MCP server. A thread on how it works 🧵👇"
X Link 2026-02-11T14:10Z [----] followers, [--] engagements

"The Problem: Staring at a blank canvas is paralyzing. You know you need a "Load Balancer" connected to [--] "Web Servers" and a "Redis Cache" but dragging them out takes time. http://draw.io http://draw.io"
X Link 2026-02-11T14:10Z [----] followers, [--] engagements

"Refinement Loop: Its not just a one-shot. You can chat with it to refine: "Add a CDN in front." "Make the database a read-replica cluster." The diagram updates instantly"
X Link 2026-02-11T14:10Z [----] followers, [--] engagements

"Privacy First: Everything runs locally. Your architecture ideas don't leave your machine. Ollama for inference Local Node.js server Standard viewer http://draw.io http://draw.io"
X Link 2026-02-11T14:10Z [----] followers, [--] engagements

"Try it out Its open source. 🔗 🔗 Blog: Contributions welcome Let me know what you think. #AI #Engineering #SystemDesign https://dev.to/harishkotra/building-an-ai-architecture-generator-with-ollama-and-drawio-mcp-4ppc https://github.com/harishkotra/SketchStack/ https://dev.to/harishkotra/building-an-ai-architecture-generator-with-ollama-and-drawio-mcp-4ppc https://github.com/harishkotra/SketchStack/"
X Link 2026-02-11T14:10Z [----] followers, [--] engagements

"Here's what I built with BulBul v3 Voice of Inclusion: A simple chrome extension that can translate and listen to text in Indian languages using @SarvamAI Demo: #TheMicIsYours Github: https://github.com/harishkotra/voice-of-inclusion https://youtu.be/IQ71xEhSC2g https://github.com/harishkotra/voice-of-inclusion https://youtu.be/IQ71xEhSC2g"
X Link 2026-02-11T18:31Z [----] followers, [--] engagements

"Day [--] of [----] building What I built on Day [--] just got a massive upgrade You can now generate hand-drawn software architecture diagrams instantly from text. Powered by @excalidraw + Local LLMs (@ollama). Day [--] of [----] building Just dropped: SketchStack An AI tool that turns your words into professional architecture diagrams in seconds. Powered by @ollama and the new @drawio MCP server. A thread on how it works 🧵👇 https://t.co/4Qqzjqa5Ix Day [--] of [----] building Just dropped: SketchStack An AI tool that turns your words into professional architecture diagrams in seconds. Powered by @ollama"
X Link 2026-02-13T07:48Z [----] followers, [--] engagements

Limited data mode. Full metrics available with subscription: lunarcrush.com/pricing