# ![@daRubberDuckiee Avatar](https://lunarcrush.com/gi/w:26/cr:twitter::1393319209367572481.png) @daRubberDuckiee Jess Wang

Jess Wang posts on X about braintrust, ai, step, code the most. They currently have [---] followers and [--] posts still getting attention that total [-----] engagements in the last [--] hours.

### Engagements: [-----] [#](/creator/twitter::1393319209367572481/interactions)
![Engagements Line Chart](https://lunarcrush.com/gi/w:600/cr:twitter::1393319209367572481/c:line/m:interactions.svg)

- [--] Week [---] -51%
- [--] Month [-----] +51%
- [--] Months [-----] +29%

### Mentions: [--] [#](/creator/twitter::1393319209367572481/posts_active)
![Mentions Line Chart](https://lunarcrush.com/gi/w:600/cr:twitter::1393319209367572481/c:line/m:posts_active.svg)

- [--] Week [--] +37%
- [--] Month [--] +356%
- [--] Months [--] +440%

### Followers: [---] [#](/creator/twitter::1393319209367572481/followers)
![Followers Line Chart](https://lunarcrush.com/gi/w:600/cr:twitter::1393319209367572481/c:line/m:followers.svg)

- [--] Week [---] +0.17%
- [--] Month [---] +3.50%
- [--] Months [---] +24%

### CreatorRank: undefined [#](/creator/twitter::1393319209367572481/influencer_rank)
![CreatorRank Line Chart](https://lunarcrush.com/gi/w:600/cr:twitter::1393319209367572481/c:line/m:influencer_rank.svg)

### Social Influence

**Social category influence**
[cryptocurrencies](/list/cryptocurrencies)  [technology brands](/list/technology-brands)  [social networks](/list/social-networks) 

**Social topic influence**
[braintrust](/topic/braintrust) #22, [ai](/topic/ai), [step](/topic/step), [code](/topic/code), [in the](/topic/in-the), [how to](/topic/how-to), [model](/topic/model), [if you](/topic/if-you), [spam](/topic/spam), [sf](/topic/sf)

**Top accounts mentioned or mentioned by**
[@braintrust](/creator/undefined) [@warpdotdev](/creator/undefined) [@cpenned](/creator/undefined) [@zachlloydtweets](/creator/undefined) [@ankrgyl](/creator/undefined) [@docker](/creator/undefined) [@joetannenbaum](/creator/undefined) [@prathkum](/creator/undefined) [@bholmesdev](/creator/undefined) [@csabakissi](/creator/undefined) [@joshwootonn](/creator/undefined) [@georgekurdin](/creator/undefined) [@usemonk](/creator/undefined) [@nateberkopec](/creator/undefined) [@morganepaloma](/creator/undefined) [@notionhq](/creator/undefined) [@aakashgupta](/creator/undefined) [@dvddkkim](/creator/undefined) [@tailwindcss](/creator/undefined) [@jjrichardtang](/creator/undefined)

**Top assets mentioned**
[Braintrust (BTRST)](/topic/braintrust)
### Top Social Posts
Top posts by engagements in the last [--] hours

"@joetannenbaum Seems like they do :)"  
[X Link](https://x.com/daRubberDuckiee/status/1745565883064590647)  2024-01-11T21:58Z [---] followers, [--] engagements


"@cpenned Congrats Chris You deserve all of this success. Can't wait to see you reach 100k and more :)"  
[X Link](https://x.com/daRubberDuckiee/status/1754002625362813312)  2024-02-04T04:43Z [---] followers, [--] engagements


"Watched a [--] min video explaining MCP. Learnings: [--]. LLMs are useless by themselves so we need to glue LLMs to a bunch of tools to make them actually useful [--]. If every tool is a different language (English Japanese Spanish) MCP is a translation layer that converts them to a universal language for the LLM to understand [--]. Nowadays its on services to construct MCP servers so that the client (Cursor Windsurf) can access the service Included the [--] most educational clips (with examples) from the video👇"  
[X Link](https://x.com/daRubberDuckiee/status/1902976957370995087)  2025-03-21T06:53Z [---] followers, [---] engagements


"2. But it becomes frustrating to glue a bunch of different tools to LLMs using different automation tools/pipelines"  
[X Link](https://x.com/daRubberDuckiee/status/1902976962450088424)  2025-03-21T06:53Z [---] followers, [--] engagements


"3. Every tool is like different language (English Japanese Spanish) and MCP is a translation layer that converts them to a universal language for the LLM to understand"  
[X Link](https://x.com/daRubberDuckiee/status/1902976965058896240)  2025-03-21T06:53Z [---] followers, [--] engagements


"4/6 Claude Code works towards full autonomy but gains your trust first"  
[X Link](https://x.com/daRubberDuckiee/status/1903304139066708484)  2025-03-22T04:34Z [---] followers, [--] engagements


"5/6 Claude Code runs terminal commands more naturally than Cursor Agent - e.g. beautiful git commit messages"  
[X Link](https://x.com/daRubberDuckiee/status/1903304141159682466)  2025-03-22T04:34Z [---] followers, [--] engagements


"I love remote work"  
[X Link](https://x.com/daRubberDuckiee/status/1904385318389981608)  2025-03-25T04:10Z [---] followers, [---] engagements


"How to set up an MCP server for beginners Using Brave MCP + Cursor Yes I deleted my API key after filming"  
[X Link](https://x.com/daRubberDuckiee/status/1908991432477847810)  2025-04-06T21:13Z [---] followers, [---] engagements


"Join @zachlloydtweets and @Prathkum is building a social listening Slackbot with @warpdotdev"  
[X Link](https://x.com/daRubberDuckiee/status/1919785309610414426)  2025-05-06T16:04Z [---] followers, [---] engagements


"I've been wanting to start a collection of educational materials at Warp for many years now but just didn't have the bandwidth. I'm so excited to see Warp University finally launch huge kudos to my team at Warp (@BHolmesDev @zachlloydtweets) for collaborating on this We've been getting a lot of questions about how to use Warp effectively to code and learn prompt-driven development. So. We launched Warp University TODAY with [--] videos: 🔥 Getting started guides 🔥 Useful developer workflows 🔥 Using MCP servers 🔥 Setting custom Rules 🔥 https://t.co/Q7GonwewMT We've been getting a lot of"  
[X Link](https://x.com/daRubberDuckiee/status/1955745870210850899)  2025-08-13T21:38Z [---] followers, [----] engagements


"@csaba_kissi @warpdotdev There are some new ones Especially under the "developer workflows" and "getting started" sections"  
[X Link](https://x.com/daRubberDuckiee/status/1955751196716884085)  2025-08-13T21:59Z [---] followers, [--] engagements


"We were quite positive about AI in this episode weirdly enough How does AI change the learning journey Ep [--] of convos.dev is out with @daRubberDuckiee https://t.co/fqE6uGKZsO How does AI change the learning journey Ep [--] of convos.dev is out with @daRubberDuckiee https://t.co/fqE6uGKZsO"  
[X Link](https://x.com/daRubberDuckiee/status/1961819977314672902)  2025-08-30T15:55Z [---] followers, [---] engagements


"1/ GitHub MCP Server Connects AI to GitHub repos for seamless code management PR reviews and issue tracking 🔗 https://github.com/github/github-mcp-server https://github.com/github/github-mcp-server"  
[X Link](https://x.com/daRubberDuckiee/status/1977158059731382531)  2025-10-11T23:43Z [---] followers, [--] engagements


"3/ Filesystem MCP Server Gives AI models secure access to your local file system for reading and writing files 🔗 https://github.com/modelcontextprotocol/servers https://github.com/modelcontextprotocol/servers"  
[X Link](https://x.com/daRubberDuckiee/status/1977158081831170068)  2025-10-11T23:43Z [---] followers, [--] engagements


"5/ Playwright MCP Server Automates browser interactions for web scraping testing and automation workflows 🔗 https://github.com/browserbase/mcp-server-browserbase https://github.com/browserbase/mcp-server-browserbase"  
[X Link](https://x.com/daRubberDuckiee/status/1977158103587004478)  2025-10-11T23:43Z [---] followers, [--] engagements


"6/ PostgreSQL MCP Server Connects AI directly to PostgreSQL databases for querying and data analysis 🔗 https://github.com/modelcontextprotocol/servers/tree/main/src/postgres https://github.com/modelcontextprotocol/servers/tree/main/src/postgres"  
[X Link](https://x.com/daRubberDuckiee/status/1977158114395742569)  2025-10-11T23:43Z [---] followers, [--] engagements


"7/ Sequential Thinking MCP Enhances AI reasoning by enabling step-by-step thinking and complex problem solving 🔗 https://github.com/sequentialread/mcp-server-sequential-thinking https://github.com/sequentialread/mcp-server-sequential-thinking"  
[X Link](https://x.com/daRubberDuckiee/status/1977158125267374469)  2025-10-11T23:43Z [---] followers, [--] engagements


"1/ GitHub MCP Server Connect AI to GitHub repos - seamless code management PR reviews & issue tracking https://github.com/github/github-mcp-server https://github.com/github/github-mcp-server"  
[X Link](https://x.com/daRubberDuckiee/status/1977158804119666972)  2025-10-11T23:46Z [---] followers, [--] engagements


"3/ Filesystem MCP Server 📁 Secure local file system access - reading & writing files made easy for AI https://github.com/modelcontextprotocol/servers https://github.com/modelcontextprotocol/servers"  
[X Link](https://x.com/daRubberDuckiee/status/1977158842048737687)  2025-10-11T23:46Z [---] followers, [--] engagements


"5/ Playwright MCP Server 🌐 Browser automation for scraping testing & workflows - web interaction simplified https://github.com/browserbase/mcp-server-browserbase https://github.com/browserbase/mcp-server-browserbase"  
[X Link](https://x.com/daRubberDuckiee/status/1977158879982075990)  2025-10-11T23:46Z [---] followers, [--] engagements


"6/ PostgreSQL MCP Server 🗄 Direct database connections - query & analyze data with AI https://github.com/modelcontextprotocol/servers/tree/main/src/postgres https://github.com/modelcontextprotocol/servers/tree/main/src/postgres"  
[X Link](https://x.com/daRubberDuckiee/status/1977158899057709118)  2025-10-11T23:46Z [---] followers, [--] engagements


"7/ Sequential Thinking MCP 🧠 Enhanced reasoning through step-by-step thinking - unlocking complex problem solving https://github.com/sequentialread/mcp-server-sequential-thinking https://github.com/sequentialread/mcp-server-sequential-thinking"  
[X Link](https://x.com/daRubberDuckiee/status/1977158917978222972)  2025-10-11T23:46Z [---] followers, [--] engagements


"@JoshWootonn @braintrust I work remote in Seattle but I'll be down in SF end of Jan :)"  
[X Link](https://x.com/daRubberDuckiee/status/2008263726584262854)  2026-01-05T19:45Z [---] followers, [--] engagements


"@GeorgeKurdin @usemonk @braintrust Hey George braintrust devrel here Great blog this is super cool learnings ❤ Curious how many / what percentage of the transactions needed to be manually reviewed by a human"  
[X Link](https://x.com/daRubberDuckiee/status/2008984062502056137)  2026-01-07T19:28Z [---] followers, [--] engagements


"I'll be attending this and helping out with one of the workshops Super excited 😁 Trace lineup is live. Speakers from Ramp Replit Notion Zendesk Dropbox HubSpot FanDuel Box and more. Bring your team. In person only. Space is limited. Feb [--] in SF https://t.co/IEI02QpLx5 https://t.co/9MaOLlF9LQ Trace lineup is live. Speakers from Ramp Replit Notion Zendesk Dropbox HubSpot FanDuel Box and more. Bring your team. In person only. Space is limited. Feb [--] in SF https://t.co/IEI02QpLx5 https://t.co/9MaOLlF9LQ"  
[X Link](https://x.com/daRubberDuckiee/status/2009077651710132358)  2026-01-08T01:40Z [---] followers, [---] engagements


"@nateberkopec Try using Braintrust to add observability to your ralphing. If you're using Claude Code: https://www.braintrust.dev/blog/claude-code-braintrust-integration https://www.braintrust.dev/blog/claude-code-braintrust-integration"  
[X Link](https://x.com/daRubberDuckiee/status/2009391665933688988)  2026-01-08T22:28Z [---] followers, [--] engagements


"@morgane_paloma @braintrust @NotionHQ @ankrgyl @aakashgupta meow :)"  
[X Link](https://x.com/daRubberDuckiee/status/2009789807694934429)  2026-01-10T00:50Z [---] followers, [---] engagements


"@dvddkkim @braintrust YEAAAAHHH WHOOOOOOOOOO"  
[X Link](https://x.com/daRubberDuckiee/status/2012342971333857448)  2026-01-17T01:55Z [---] followers, [---] engagements


"Not great at responding to comment on my channels but trying to make it more of a habit. Here's some noteworthy ones: 🔵 OpenCode as Claude Code alternative 🔵 Workbeaver instead of Claude Cowork 🔵 Workmux or Conductor for orchestrating agents 🔵 Gamma AI for slide creation 🔵 Of course obligatory shoutout to @warpdotdev in the comments 😁 & these go on a (impossibly) growing list of tools to try for when I get time to try new things. https://twitter.com/i/web/status/2013627587457658883 https://twitter.com/i/web/status/2013627587457658883"  
[X Link](https://x.com/daRubberDuckiee/status/2013627587457658883)  2026-01-20T15:00Z [---] followers, [---] engagements


"Been trying to find a second brain setup that works for me: 🔵 on my phone to take voice notes and transcribe them 🔵 Go onto the desktop app and download these transcriptions (1 by [--] because I'm not paying for the paid version) 🔵 When I want to do second brain stuff zip up the transcriptions 🔵 Pull the .zip into ChatGPT which is I've also connected to my Notion Gmail Google Calendar. 🔵 Pull in a prompt for the synthesis layer (brainstorming to-do's etc). Prompts here 👉 Right now it's still finnicky. Not sure if it's the prompts' fault or just the workflow itself."  
[X Link](https://x.com/daRubberDuckiee/status/2014714747166588981)  2026-01-23T15:00Z [---] followers, [--] engagements


"DAY [--] OF NEW JOB @braintrust"  
[X Link](https://x.com/daRubberDuckiee/status/2008235961357099127)  2026-01-05T17:55Z [---] followers, [---] engagements


"Respect to @braintrust for sponsoring @tailwindcss ❤"  
[X Link](https://x.com/daRubberDuckiee/status/2010970065244733598)  2026-01-13T07:00Z [---] followers, [---] engagements


"If you're building an app with AI it's important to be able to optimize your prompt and log the cost and latency of your LLM logs. I made a really compact [--] minute demo that covers a lot of the basics like playgrounds experiments prompt versioning testing in CI/CD etc"  
[X Link](https://x.com/daRubberDuckiee/status/2015801908641092038)  2026-01-26T15:00Z [---] followers, [---] engagements


"shwaaaagggg @braintrust ❤"  
[X Link](https://x.com/daRubberDuckiee/status/2016164299644010821)  2026-01-27T15:00Z [---] followers, [---] engagements


"I was thinking which models write more "annoying" E.g. using the common AI-isms like excessive em dashes "it's not X it's Y" etc. Built an eval in [--] min to find out:"  
[X Link](https://x.com/anyuser/status/2016424327722491933)  2026-01-28T08:13Z [---] followers, [---] engagements


"Step #1: dataset Had Loop in @braintrust generate [--] writing prompts of various topics to get a range of outputs"  
[X Link](https://x.com/daRubberDuckiee/status/2016424329408610522)  2026-01-28T08:13Z [---] followers, [--] engagements


"Step #2: task + scorer. Task was simple just pass in the writing prompt we just generated. For scoring I asked Loop to write an LLM-as-judge for me. And had to tweak it from a flat count of AI-isms to percentage proportional to number of words in the output"  
[X Link](https://x.com/daRubberDuckiee/status/2016424330901835967)  2026-01-28T08:13Z [---] followers, [--] engagements


"Link to the repo is in the thread if you want grab the dataset / code for yourself to try Your agent can be better at detecting spam. Here's how to write an eval to compare o3-mini to Claude [---] Sonnet on spammy messages and why the lower-scoring model might actually be right. Your agent can be better at detecting spam. Here's how to write an eval to compare o3-mini to Claude [---] Sonnet on spammy messages and why the lower-scoring model might actually be right"  
[X Link](https://x.com/anyuser/status/2017296834625175829)  2026-01-30T18:00Z [---] followers, [---] engagements


"@braintrust https://github.com/braintrustdata/braintrust-cookbook/tree/main/examples/SpamClassifier https://github.com/braintrustdata/braintrust-cookbook/tree/main/examples/SpamClassifier"  
[X Link](https://x.com/daRubberDuckiee/status/2017303754039783485)  2026-01-30T18:27Z [---] followers, [--] engagements


"Somebody commented on my video: "Ralph Wiggum is just a while loop for LLMs" and I thought that was a very apt description of what may seem like a complicated concept"  
[X Link](https://x.com/anyuser/status/2017432659358847420)  2026-01-31T03:00Z [---] followers, [---] engagements


"@jjrichardtang @rootlyhq @mintlify @braintrust @handotdev @ankrgyl Seahawks for sure"  
[X Link](https://x.com/daRubberDuckiee/status/2018830525419651554)  2026-02-03T23:34Z [---] followers, [--] engagements


"Sometimes I forget how big of a step function up certain models are. Here's a Mermaid CLI diagram created by GPT-4o versus Opus [---] (I don't have you tell you which one is which)"  
[X Link](https://x.com/anyuser/status/2021339402606346247)  2026-02-10T21:44Z [---] followers, [---] engagements


"@mikepmunroe Braintrust devrel here that's awesome to hear. Let us know if you have any cookbooks/use cases you want us to add"  
[X Link](https://x.com/daRubberDuckiee/status/2022359813263692158)  2026-02-13T17:18Z [---] followers, [--] engagements


"Gonna be up in SF Jan 29th to do a demo with Braintrust on AI evals 💪. Drop by if you're in the area https://luma.com/evals_on_tap_sf https://luma.com/evals_on_tap_sf"  
[X Link](https://x.com/daRubberDuckiee/status/2010890337896841446)  2026-01-13T01:43Z [---] followers, [---] engagements


"Let AI code for you while you sleep. ok fine 🫣 .but here's how to Braintrust to keep an eye on cost & errors in case things don't go according to plan. https://www.braintrust.dev/blog/ralph-wiggum-debugging My Ralph Wiggum breakdown went viral. It's a keep-it-simple-stupid approach to AI coding that lets you ship while you sleep. So here's a full explanation example code and demo. https://t.co/FyVdrIyqUP https://www.braintrust.dev/blog/ralph-wiggum-debugging My Ralph Wiggum breakdown went viral. It's a keep-it-simple-stupid approach to AI coding that lets you ship while you sleep. So here's"  
[X Link](https://x.com/daRubberDuckiee/status/2011236577968885777)  2026-01-14T00:39Z [---] followers, [---] engagements


"@sarahzengy @braintrust this is so cute lol"  
[X Link](https://x.com/daRubberDuckiee/status/2022382850214105590)  2026-02-13T18:50Z [---] followers, [--] engagements


"Take human review on the go. Human annotation and scoring is now optimized for mobile in Braintrust"  
[X Link](https://x.com/anyuser/status/2022764169695551859)  2026-02-14T20:05Z [----] followers, [---] engagements


"new podcast episode dropped with @cpenned thumbnail says it all 🙂 https://www.youtube.com/watchv=vf5Q-pxf7iE https://www.youtube.com/watchv=vf5Q-pxf7iE"  
[X Link](https://x.com/anyuser/status/2022324889168826527)  2026-02-13T15:00Z [---] followers, [---] engagements


"Ive been reflecting on my relationship with content so I wanted to share a short story about how I got here. Ive always been a huge consumer of content. I grew up watching gaming channels like Rooster Teeth and Day9's StarCraft streams followed creators like David So Nigahiga and Wong Fu and watched a lot of anime and movies. I didnt have the happiest time in high school or college and those creators had a bigger impact on me than I realized at the time. In college and a few years into working full time I filmed a handful of YouTube videos and even made a small documentary about an arcade"  
[X Link](https://x.com/anyuser/status/2021962502167314480)  2026-02-12T15:00Z [---] followers, [---] engagements


"I feel like the ability to multitask with agents in parallel is a trap. Last week I found myself working on four pretty beefy tasks at the same time. On paper that sounds great but I actually ended the week feeling unsatisfied. Every time I switched contexts I had to re-establish where I was: What was this agent doing What did I miss while I was gone There was also this weird background anxiety of trying to get everything done which meant I was rushing to skim output and code (not fully reading or understanding it) just so I could unblock the next parallel agent as quickly as possible. Theres"  
[X Link](https://x.com/anyuser/status/2021600112737439858)  2026-02-11T15:00Z [---] followers, [---] engagements


"Sometimes I forget how big of a step function up certain models are. Here's a Mermaid CLI diagram created by GPT-4o versus Opus [---] (I don't have you tell you which one is which)"  
[X Link](https://x.com/anyuser/status/2021339402606346247)  2026-02-10T21:44Z [---] followers, [---] engagements


"Super excited about being able to use it on my phone 👀 Introducing Oz: the platform to orchestrate agents in the cloud. Spin up hundreds of agents from your terminal browser the API or your phone. Each agent gets a @docker environment to build test and write PRs. Come back from your lunch break to code thats ready to merge. https://t.co/gdHjcwaU4q Introducing Oz: the platform to orchestrate agents in the cloud. Spin up hundreds of agents from your terminal browser the API or your phone. Each agent gets a @docker environment to build test and write PRs. Come back from your lunch break to code"  
[X Link](https://x.com/anyuser/status/2021308224100532419)  2026-02-10T19:40Z [---] followers, [---] engagements


"Introducing Oz: the platform to orchestrate agents in the cloud. Spin up hundreds of agents from your terminal browser the API or your phone. Each agent gets a @docker environment to build test and write PRs. Come back from your lunch break to code thats ready to merge"  
[X Link](https://x.com/anyuser/status/2021247666965958999)  2026-02-10T15:39Z 47.1K followers, 2.3M engagements


"AI image generators are notoriously bad at generating text. Here's why. These models use a process called diffusion. They start with pure noise and gradually remove it step by stepessentially guessing which pixels should be dark versus light based on patterns they've seen before. This works great for organic shapes like faces or trees. Small imperfections don't matter. A face that's slightly off still looks like a face. But text is unforgiving. The difference between O and Q is a single line. The letters RN next to each other look a lot like M. These tiny distinctions are easy for humans to"  
[X Link](https://x.com/anyuser/status/2021237734657097979)  2026-02-10T15:00Z [---] followers, [--] engagements


"What's new: - Spans as table rows: filter experiments at the span level - Autocomplete + linting: validate your code when writing scorers - Data point comparison: see how experiments performed on one input - Multiple trace views: switch between custom views for easier review"  
[X Link](https://x.com/anyuser/status/2020914470864785504)  2026-02-09T17:35Z [----] followers, [---] engagements


"Evaluate how good Twelve Labs' language model Pegasus is at actually understanding what's going on in a video using Hugging Face's MMVU dataset & Braintrust for evals"  
[X Link](https://x.com/anyuser/status/2019861295072194939)  2026-02-06T19:50Z [----] followers, [---] engagements


"Anybody else name their experiments like this when they're trying to get their evals working"  
[X Link](https://x.com/anyuser/status/2019676360978186556)  2026-02-06T07:35Z [---] followers, [---] engagements


"I know it's not a big number but I should be hitting 10k subscribers on Youtube soon which was my goal all the way at the beginning of [----] 🙂"  
[X Link](https://x.com/anyuser/status/2019264626249330784)  2026-02-05T04:19Z [---] followers, [---] engagements


"Somebody commented on my video: "Ralph Wiggum is just a while loop for LLMs" and I thought that was a very apt description of what may seem like a complicated concept"  
[X Link](https://x.com/anyuser/status/2017432659358847420)  2026-01-31T03:00Z [---] followers, [---] engagements


"Link to the repo is in the thread if you want grab the dataset / code for yourself to try Your agent can be better at detecting spam. Here's how to write an eval to compare o3-mini to Claude [---] Sonnet on spammy messages and why the lower-scoring model might actually be right. Your agent can be better at detecting spam. Here's how to write an eval to compare o3-mini to Claude [---] Sonnet on spammy messages and why the lower-scoring model might actually be right"  
[X Link](https://x.com/anyuser/status/2017296834625175829)  2026-01-30T18:00Z [---] followers, [---] engagements


"Your agent can be better at detecting spam. Here's how to write an eval to compare o3-mini to Claude [---] Sonnet on spammy messages and why the lower-scoring model might actually be right"  
[X Link](https://x.com/anyuser/status/2017266604073951256)  2026-01-30T16:00Z [----] followers, [---] engagements


"Was a dope event evals are a critical step in shipping quality AI. And builders are paying attention cos quality doesnt improve by accident. @braintrust evals on tap SF. https://t.co/XT7p9S9rZa evals are a critical step in shipping quality AI. And builders are paying attention cos quality doesnt improve by accident. @braintrust evals on tap SF. https://t.co/XT7p9S9rZa"  
[X Link](https://x.com/anyuser/status/2017292805086003207)  2026-01-30T17:44Z [---] followers, [--] engagements


"evals are a critical step in shipping quality AI. And builders are paying attention cos quality doesnt improve by accident. @braintrust evals on tap SF"  
[X Link](https://x.com/anyuser/status/2017271548663771536)  2026-01-30T16:19Z [---] followers, [----] engagements


"evals evals evals"  
[X Link](https://x.com/anyuser/status/2017073491196203277)  2026-01-30T03:12Z [---] followers, [----] engagements


"How I make my AI-generated images a little less annoying"  
[X Link](https://x.com/anyuser/status/2016889068345561417)  2026-01-29T15:00Z [---] followers, [--] engagements


"I was thinking which models write more "annoying" E.g. using the common AI-isms like excessive em dashes "it's not X it's Y" etc. Built an eval in [--] min to find out:"  
[X Link](https://x.com/anyuser/status/2016424327722491933)  2026-01-28T08:13Z [---] followers, [---] engagements


"Results: - Claude Sonnet [---] uses these patterns 28% of the time. - GPT-5 uses them 7.8% of the time"  
[X Link](https://x.com/anyuser/status/2016424334701822202)  2026-01-28T08:13Z [---] followers, [---] engagements


"And yes I know [--] rows of data isn't enough to feel confident and we should test with different models. But that's not the point. The point is that evals don't have to be perfect to be useful. You have a question you write a test you get signal. This took [--] minutes on a plane (and plane wifi) 🙂 If you're not writing evals because it feels too formal or slow you're overthinking it"  
[X Link](https://x.com/anyuser/status/2016424336681549831)  2026-01-28T08:13Z [---] followers, [--] engagements

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