[GUEST ACCESS MODE: Data is scrambled or limited to provide examples. Make requests using your API key to unlock full data. Check https://lunarcrush.ai/auth for authentication information.]

@sergebevz Avatar @sergebevz Serge Bevzenko

Serge Bevzenko posts on X about ai, $googl, open ai, apple the most. They currently have XX followers and XX posts still getting attention that total XXX engagements in the last XX hours.

Engagements: XXX #

Engagements Line Chart

Mentions: XX #

Mentions Line Chart

Followers: XX #

Followers Line Chart

CreatorRank: XXXXXXXXX #

CreatorRank Line Chart

Social Influence

Social category influence technology brands XXXXX% stocks XXXXX% finance XXXX% vc firms #339 celebrities XXXX% social networks XXXX%

Social topic influence ai 25.93%, $googl 5.56%, open ai 5.56%, apple 5.56%, $475m 3.7%, money 3.7%, the first 3.7%, meta 3.7%, solve 1.85%, $newr XXXX%

Top assets mentioned Alphabet Inc Class A (GOOGL) New Relic, Inc. (NEWR) Microsoft Corp. (MSFT) IBM (IBM)

Top Social Posts

Top posts by engagements in the last XX hours

"AI datacenters will hit an energy wall in 3-4 years. One startup just raised $475M in a SEED round to solve this. Here's what Unconventional AI is building and why Bezos a16z and Sequoia are betting on it:"
X Link 2025-12-11T09:36Z XX followers, XX engagements

"The problem: Datadog and New Relic don't cover LLM specifics. You need different tools: - LangSmith (LangChain ecosystem) - Langfuse (open source) - Arize Phoenix (RAG agents)"
X Link 2025-12-08T14:42Z XX followers, XX engagements

"54% of companies report "AI monitoring adoption." But that's AI inside observability smarter alerts auto root cause analysis. Not observability for AI. Like confusing a hammer with a microscope"
X Link 2025-12-08T14:42Z XX followers, XX engagements

"Already tested on: Language modeling DNA sequence analysis Time series forecasting Architecture generalizes beyond just text"
X Link 2025-12-08T23:12Z XX followers, XX engagements

"Benchmarks don't lie: MMMLU (8 languages): XXXX (beats DeepSeek's 84.2) GPQA-Diamond: XXXX (beats DeepSeek's 41.9) #2 among open-source models on LMArena This is frontier-level performance. For free"
X Link 2025-12-09T07:26Z XX followers, XX engagements

"The bigger picture: X years ago frontier AI = OpenAI monopoly X year ago Anthropic and Google joined Now open source is catching up Apache XXX means anyone can use modify commercialize. No API lock-in. No usage restrictions"
X Link 2025-12-09T07:26Z XX followers, XX engagements

"A week ago I posted about OpenAI's "Code Red." Here's what happened since: GPT-5.2 rushed out X weeks early 12M daily users gone A suspicious "productivity report""
X Link 2025-12-09T11:46Z XX followers, XXX engagements

"So which is it Panic and user exodus Or 320x growth in advanced AI usage Maybe both. The AI race is now too fast to know who's winning"
X Link 2025-12-09T11:51Z XX followers, XX engagements

"Sources: GPT-5.2 acceleration: 12M users: MIT vs OpenAI report: 320x reasoning:"
X Link 2025-12-09T11:51Z XX followers, XXX engagements

"The human brain processes incredible amounts of information on XX watts. A single AI datacenter Hundreds of MEGAWATTS. Unconventional AI wants to build chips that work "like biology.""
X Link 2025-12-11T09:36Z XX followers, X engagements

"The money is insane. $475M seed round at $4.5B valuation. For context: typical seed is $1-5M. This is just the FIRST tranche they plan to raise up to $1B total"
X Link 2025-12-11T09:37Z XX followers, X engagements

"OpenAI's response Created the Frontier Risk Council. Built Aardvark a security scanning tool. Published safety guidelines. All good steps"
X Link 2025-12-11T16:39Z XX followers, XX engagements

"But here's the thing. In July they released ChatGPT Agent. It already rated "high" on their own risk framework. They released it anyway"
X Link 2025-12-11T16:39Z XX followers, X engagements

"ARC Prize 2025 results just dropped. A 4B model from NVIDIA beat systems 100x larger. Cost per task: $XXXX vs $XX for frontier models. The secret isn't size. It's test-time training"
X Link 2025-12-10T08:04Z XX followers, XX engagements

"NVIDIA's NVARC team (two Kaggle Grandmasters) hit 27.64%. Their approach: Generate synthetic training data with open-weight models Train the model ON EACH PUZZLE during inference No chain-of-thought no massive compute The model learns while being tested"
X Link 2025-12-10T08:05Z XX followers, XX engagements

"But here's where it gets wild. Samsung's TRM: 7M parameters. Scored XX% on ARC-AGI-1. That's XXXX% the size of GPT-4. Beating most frontier models. How Recursive self-improvement XX iterations of refining its own answer"
X Link 2025-12-10T08:05Z XX followers, XX engagements

"Also connects to Apple's "Illusion of Thinking" debate. They said reasoning models collapse on complex tasks. ARC Prize shows: yes static models do. But models that adapt at test-time don't"
X Link 2025-12-10T08:08Z XX followers, XX engagements

"The founder is Naveen Rao. His track record: - Founded Nervana Systems sold to Intel for $350M+ - Founded MosaicML sold to Databricks for $1.3B - PhD in neuroscience from Brown - Engineering degree from Stanford This is his 3rd AI hardware company"
X Link 2025-12-11T09:36Z XX followers, XX engagements

"Target models: diffusion flow models energy-based models. These are inherently non-deterministic they naturally fit analog computing's nonlinear dynamics. Not trying to replace ALL GPUs. Targeting specific workloads"
X Link 2025-12-11T09:37Z XX followers, X engagements

"The investors list reads like a VC hall of fame: - Andreessen Horowitz (lead) - Lightspeed (lead) - Sequoia - Lux Capital - DCVC - Future Ventures - Jeff Bezos ($10M personal) - Rao himself ($10M of his own money)"
X Link 2025-12-11T09:37Z XX followers, XXX engagements

"But here's the context they don't emphasize: Big Tech lobbied hard. Meta Google Amazon Apple pushed back against strict rules warning AI development would move outside Europe. CCIA (tech lobby) welcomed the delay but wanted "bolder" deregulation"
X Link 2025-12-11T13:01Z XX followers, XX engagements

"The Draghi report looms large. Former ECB chief Mario Draghi warned that EU's XXX tech laws and 270+ digital regulators are killing competitiveness. His finding: XX% of the EU-US GDP gap comes from lower productivity"
X Link 2025-12-11T13:02Z XX followers, XX engagements

"42 US state attorneys general just sent a letter to XX AI companies. The message: fix your "delusional" chatbots or face legal consequences. This is not a request. It's an ultimatum"
X Link 2025-12-11T20:14Z XX followers, XX engagements

"Who received the letter: OpenAI Google Microsoft Meta Apple Anthropic xAI Perplexity Character AI Chai AI Replika Nomi AI. Basically everyone building consumer chatbots"
X Link 2025-12-11T20:14Z XX followers, XX engagements

"The past year brought a series of tragedies linked to AI chatbots. The victims: Sewell Setzer (14) Amaurie Lacey (17) Zane Shamblin (23) Suzanne Adams (83). Suicides. And the first murder case connected to a chatbot"
X Link 2025-12-11T20:14Z XX followers, XX engagements

"IBM just dropped $11B on Confluent. Biggest acquisition in years. Here's what they bought and why it matters for AI:"
X Link 2025-12-12T08:26Z XX followers, XX engagements

"Coder just shipped AI Development Infrastructure a platform for hybrid human-agent teams. Finally a serious answer to "how do we let AI agents into our codebase safely" Here's what they built and why it matters:"
X Link 2025-12-12T11:23Z XX followers, XX engagements

"The problem: AI agents are getting powerful (Claude Code Cursor Devin). But enterprises are stuck. Developers run agents locally share API keys in Slack and nobody knows the real AI spend. Only XX% of companies have a clear AI strategy (Cisco)"
X Link 2025-12-12T11:23Z XX followers, XX engagements

"Agent Boundaries: policy-driven control for agents. Explicit allowlists for network destinations tools internal systems. Your AI agent physically cannot access production databases or internal APIs unless explicitly permitted"
X Link 2025-12-12T11:23Z XX followers, X engagements

"Coder Tasks: long-running automation for both humans and agents. Code review documentation issue resolution test generation all with full API and notifications. The execution layer for hybrid workflows"
X Link 2025-12-12T11:23Z XX followers, X engagements