@martinfowler.com do you want to keynote our next virtual conference on AI agents in production? home.mlops.community/public/event...

Thanks @mlopscommunity.bsky.social for the shoutout in today's newsletter! www.controlpla.in/blog/prompt-...

Learning from Raza about how he thinks about evals and reliable AI

I am hosting this one next week. it's virtual and free with a bunch of great speakers in case you are free drop by! And I'll be playing some live music during the breaks! www.openxdata.ai

I vibed too hard and ended up creating a community and merch around VibeOps. Do I regret it...? So far sales have been pretty good. I'm about to set up a reddit AMA on how to win at Vibe drop shipping.

Whenever I talk with someone about @modal-labs.bsky.social they mention how it reinvented the developer experience. CEO Erik Bernhardsson does hard things. Like wrangle GPUs. Come hear his final keynote - home.mlops.community/home/events/...

putting AI into production at the next virtual conference on March 12th. Come join us home.mlops.community/home/events/...

another agent hour happening tonight hearing from folks building agents and what they have learned home.mlops.community/home/events/...

@jeffdean.bsky.social want to keynote our virtual conference on March 12th? home.mlops.community/home/events/...

I want to organize a debate on Web Agents vs API agents. Let me know if you want to argue one side or the other.

Tech debt for data teams is not refactoring code or having to update libraries. Tech debt for data teams is figuring out which of the 11,000 dbt pipelines are actually being used

Nobody talking about how Kokoro is raising the bar for open-source TTS while Elevenlabs is just out here raising their valuation. But yah, deepseek is cool.

Robert was an absolute gem on this one being super transparent about whats working for him building knowledge graphs

Hosting another agent hour tomorrow. come join us home.mlops.community/public/event...

This is the third time in the last 2 months I have heard a guest talk about how our biggest bottleneck working with large models is the memory constraints. All from people I would consider to be preeettyyyy legit Here is a clip taken from my conversation with Krishna Sridhar from Qualcomm

AWS Lambda functions have an incredible opportunity to rebrand as agents

another agent hours happening this week. got some great speakers and roundtable sessions lined up. home.mlops.community/home/events/...

Key Insights from the YouGot.us 2024 AI Agents Survey - Use cases: Coding, content generation, and process automation. - Frustrations: Lack of reliability and hallucinations remain major pain points, along privacy concerns. yougot.us/posts/2024-1... #MLOps #AIAgents @mlopscommunity.bsky.social

if you are fine-tuning models hit me up. I am organizing small group sessions with NVIDIA to understand 1. What tools are useful 2. How the ecosystem around fine-tuning could be improved

building an ML platform from scratch live streaming now youtube.com/live/9LxsViI...

is anyone building mixpanel for AI agent use? Like which agents are being used and effective, how they're doing on their tasks etc. I guess its a form of evaluation but more like evaluating the pmf of the agent and less the output. (out eval also good)

@pydantic.dev AI talk with Samuel Colvin (creator) tomorrow. Come hang out home.mlops.community/public/event...

Probably the Gemini family of models. And biggest surprise of the year would be Qwen-2.5

Creating AI awards of 2024. So far the categories I have are: - Most censored model: the original Gemini? - Best aquahire: Adept? Character AI? Inflection? - Best model thats bad at benchmarks: Claude Sonnet? - Best/Worst AI Product: Truth terminal? - Best open source model: llama 3.2? What else?

I'm with Eric Riddoch and Ben Epstein live coding with SQLmesh. Come join us youtube.com/live/6Ivmza-...

All of these companies exist. Except one. 2024 was the year of the Lang - Langchain - Langtrace - Langwatch - Langchung - Langfuse - Langflow

whats your prediction for 2025? (wrong answers only)

Agents are to tools what ML Models are to Jupyter Notebooks. Ok. I know. That doesn't make much sense. But stay with me. Tool Calling gives AI power. It allows LLMs to interact with external systems and perform tasks dynamically, which is kinda important for building agents.

For all the folks at the crossroads of Excel & DuckDB, this just went open source

if this is you send me a DM so i can send you some new Merch we just created

true test of the ROI of AI for public companies. When you announce your AI driven does the stock price go up more than the LLM API calls cost

We are having our first 'agent hours' round table sessions on Thursday. The conversations will center around these topics: - Cost Optimization Strategies - Agent UX - Decomposing Tasks With Experts - Evals for Voice Agents - Agent Auth There are 5 spots available for each topic

How will AI agents will look in the future? Will it be dominated by agents that control your computer and mouse or will it be API driven? What are the pros and cons of each?

I would like to start MLOps Community working groups next year but since i have never done that or participated in a wg before I would love some help. Thoughts on: What makes a good working group? What are key things we should keep in mind? What are some ways working groups fail?

And, when that relative asks you to build a website for their crystal chakra alignment business just because you work in tech, point them to this gem and tell them to do it themselves! I started memeing more in our weekly newsletter, readers be warned: gatewaze.mlops.community?mode=newslet...

Things I would like to know about folks doing AI these days... - What does the day-to-day entail? (managing k8s clusters and inference pipelines? Prompt tuning? Optimal chunk sizes?)

Nontechnical user adoption for ai agents is really hard right now. I spoke with a team recently who has gone all in on agents. I wanted to understand the rough edges. Here are some: **End user adoption** - This may be the biggest hurdle because:

Finally finished reading through the State of AI Report 2024. Here is what stood out to me (non-AI generated takeaways) 1. 200+ slides - Like in previous years there is no email capture nor asking for sign-up, despite tens of thousands of readers.

I really want to know who is actually using AI Agents in production. So i am organizing this to learn from the best of them. home.mlops.community/home/events/...