Sam Flamini

Applied AI

[Mike Krieger @mikeyk]

Our Applied AI team with a great write up on improving Claude’s frontend design via Skills. Also with a Claude Code plugin that packages up the skill!
https://t.co/5gPFn5UwhW

[jeremy @jerhadf]

it was awesome to work on this system, and I'm excited to share some of the things we learned! agents go brr

[David Soria Parra @dsp_]

We finalized a new revision of MCP. Revision 2025-03-26 will bring Auth, Streamable HTTP, Audio modality and a few other goodies.

We will be getting the SDKs up-to-date asap and will work towards a v 2.0 of the Python and Typescript SDK.

But don't worry, everything is versioned so your existing MCP servers will work.

[Sam Flamini @sflamini5]

@aiDotEngineer @MaheshMurag @AnthropicAI Mahesh 👑

[AI Engineer @aiDotEngineer]

The highly-anticipated MCP workshop with @MaheshMurag is now live!

Join @AnthropicAI Applied AI Engineer (+ creator of MCP) @MaheshMurag in this full 2-hour workshop covering the Model Context Protocol, where he covers:

00:00 Introduction
00:35 What is MCP?
09:39 Building with MCP
26:25 MCP & Agents
1:13:15 What's next for MCP?

YT version in next post.

[Pratik Bothra @pratik60]

Some customers see upwards of 80% resolution rate, but on AVG we see 51% resolution rate. I think that's pretty incredible!

I also want to give a special shout out to @PedroTabacof for leading the work on our end and @sflamini5 from Anthropic who has been excellent to work with.

[Sam Flamini @sflamini5]

@pratik60 @PedroTabacof Thank you Pratik! It's been a joy to work with you and the team

[Alex Albert @alexalbert__]

Excited to share our latest research on Contextual Retrieval - a technique that reduces incorrect chunk retrieval rates by up to 67%.

When combined with prompt caching, it may be one of the best techniques there is for implementing retrieval in RAG apps.

Let me explain:

[Sam Flamini @sflamini5]

Today was a special day to be @AssemblyAI 💙

[AssemblyAI @AssemblyAI]

Introducing Universal-1, our most powerful speech recognition model to date.

Trained on over 12.5 million hours of multilingual audio data, Universal-1 achieves best-in-class speech-to-text accuracy across English, Spanish, French, and German.

https://t.co/xgrE6EhmJu

[AssemblyAI @AssemblyAI]

Introducing Universal-1, our most powerful speech recognition model to date.

Trained on over 12.5 million hours of multilingual audio data, Universal-1 achieves best-in-class speech-to-text accuracy across English, Spanish, French, and German.

https://t.co/xgrE6EhmJu

[Sam Flamini @sflamini5]

Underappreciated aspect of emerging AI market:

Buying AI models is very different from buying SaaS

-your purchase decision may impact COGS directly
-pricing is based on usage, which you need to accurately forecast up front
-technical evals more important than ever when buying

[Sam Flamini @sflamini5]

@naklecha @aaaaaaaaaaorg we both work in ai now sir

[Sam Flamini @sflamini5]

Almost every LLM eval system I've ever seen/built started with google sheets

[Sam Flamini @sflamini5]

Everyone wants to build smarter AI, but maybe the real opportunity is in using AI to create smarter humans

[Sam Flamini @sflamini5]

@thatguyintech this seems to be the path of most successful 'hard tech' founders

[Sam Flamini @sflamini5]

Market risk vs technical risk

Do you want to fight competition, or physics?

[Sam Flamini @sflamini5]

@thatguyintech Steak & eggs was my go to breakfast of choice in college

[Sam Flamini @sflamini5]

@thatguyintech @deformapp oh yes. will dm

[Sam Flamini @sflamini5]

Technology often has such a high degree of specialization that going *just deep enough* into an area can quickly put you in the top 1% of domain knowledge in that space

Many engineers dramatically underestimate the relative depth of their knowledge on their favorite topics

[Sam Flamini @sflamini5]

@thatguyintech integrating AI into products & business workflows