Nabeel Hyatt

@kevinakwok ?

There should be so many more SAFE-esque structures

@claymfisher @latent_health @sparkcapital 🚀

@latent_health is one of the biggest and most strategic opportunities in healthcare, sitting in between providers, payors, patients, and pharma and $200B of pharma spend. @sparkcapital is delighted to support Rish and Sri and this consequential company and to work with @saranormous and Mike Dixon.

@billclerico @cjpedregal Good idea.

@cjpedregal I want to be able to create an account for an agent they can access via MCP and share notes with them Right now you can only access notes you own But I don’t want the agent to have access to all of my notes - just team meetings

There are some tweets out there saying that Granola is trying to lock down access to your data. Tldr; we are actually trying to become more open, not closed. We’re launching a public API next week to complement our MCP. Read on for context. A couple months ago, we noticed that some folks had reversed engineered our local cache so they could access their meeting data. Our cache was not built for this (it can change at any point), so we launched our MCP to serve this need. The MCP gives full access to your notes and transcripts (all time for paid users, time restricted for free users). MCP usage has exploded since launch, so we felt good about it. A week ago, we updated how we store data in our cache and broke the workarounds. This is on us. Stupidly, we thought we had solved these use cases well enough with our MCP. We’ve now learned that while MCPs are great for connecting to tools like Claude or chatGPT, they don’t meet your needs for agents running locally or for data export / pipeline work. So we’re going to fix this for you ASAP. First, we’ll launch a public API next week to make it easier for you to pull your data. Second, we’ll figure out how to make Granola work better for agents running locally. Whether that’s expanding our MCP, launching a CLI, a local API, etc. The industry is moving quickly here, so we’d appreciate your suggestions. We want Granola data to be accessible and useful wherever you need it. Stay tuned.

@kimmaicutler @krishnanrohit oh come on now 🥹

@krishnanrohit I know I’m going to start sobbing every time I hear the “Up” montage music.

This is exceptionally good.

@metapreston x.com/nabeel/status/…

Steve Jobs refusing to be introspective in a 2006 interview: "I don't think that way"

I literally have Granola hooked to Claude and my CRMs so I'm unsure what you mean here. I think people like the idea of access to every transcript, (which is cool Granola will allow it) but I personally think its just a dumb safety net that is notably worse at the job than just letting Granola do it. Insight extraction from reams of transcripts is actually really hard to do well. One reason Granola notes are better is because they are just better at extracting insights than the random stuff someone built last month to mimic Granola. I know because I tried handling all the transcripts myself first, and eventually realized it's better to just ping Granola to search for what you need with MCP/API than roll your own subpar stuff. But to each their own everyone is on the journey.

@nabeel think folks would not be as upset if the MCP was sufficient skills and notes are much more valuable in the place people actually work - claude, CRMs granola is a generational team and the best in-meeting product - that surface area seems defensible and still underdeveloped

granola should just be honest “it is a difficult time to build a focused product like granola. if we let people access data freely, we risk becoming an undifferentiated data collector as people access notes in apps outside of granola we need time to build beyond note taking”

@cynthiamcgillis @Shpigford @cjpedregal Yes. Transcripts are in MCP. I've found Granola is an awful lot better at summarizing and working with full transcripts than anything you can vibecode though. So I built transcript stuff, then just went back to querying Granola directly for what I needed.

@Shpigford @cjpedregal Does the MCP give full transcripts? I'm using the AI summaries rn but would prefer full transcripts. But Claude told me that would be token heavy and to start with summaries.

"The best people in computers would have been poets and writers and musicians. They went into computers because it was so compelling, because it was fresh and a new of expression for their talents... this is becoming less true unfortunately." youtu.be/cBk4a_uOi7Q?si…

Love this video of Steve Jobs being introspective of his management style in early Apple and what he's learned. Prior to his triumphant return to Apple a few years later. youtu.be/AKhGVG2BxXY?si…

This is the most important point. There is nothing, in essence, you would say about AI you would not have also said about the computer or the engine. But deciding to defacto be anti-engine or anti-computer as a culture would so deeply problematic for that culture in the following decades.

I don’t think there’s ever been en a technology whose builders constantly promise that, if they succeed, tens of millions of jobs will be destroyed and the world might end.

@appenz @meetgranola Granola did not go closed. Unsecured transcripts were obviously an issue. FWIW - the MCP for @meetgranola is awesome, I reworked all my workflows to use it first week is was out. But maybe you just want to use the API instead.

Sorry to see Granola @meetgranola going closed. They encrypted their local db, no local and no cloud API. In a world where notes are managed by agents, the app now has zero value. Any recommendations for good alternatives? What are you switching to?

@far33d @appenz @meetgranola Lemme help. "Hey Claude - can you re-write my backend that uses the unsecure local Granola database to use the secure Granola MCP instead - here are the docs docs.granola.ai/help-center/sh…"

@appenz @meetgranola This completely nuked workflows and tools I'd built on top of it for my personal CRM.

It's a narrative problem *first* because the civil & private institutions cannot mandate people to make a transition they are fundamentally resisting. Even with an excited populous adopting the computer, we needed to educate and support people in re-education and re-employment through that transition. That was with it being seen positively. If it is treated with negativity, fear, "you dont use AI right?" kind of social pressure - that will set the transition back even further. I agree that national scale coordination will be needed. But AI is an incredible job creation engine and engine of opportunity that is being meet with worry in the U.S. like no other place on the planet. That's a cultural issue first, or we'll get the policy and coordination all wrong, and people will resist the change no matter what we do.

Nvidia is worth $3T with 30,000 employees. Ford at peak employed 400,000. That ratio is what’s different. Jevons probably holds , demand for skilled work will expand as it gets cheaper. I am very optimistic about this future. But that expansion takes time. The transition is faster than any prior wave and the people who get stranded in it don’t have a next rung, yet. This isn’t a startup problem. Or even a tax problem. It’s a coordination and product problem between civil and private institutions at a national scale. Treating it as a narrative problem means people already decided the stranded people aren’t our problem.

AI is deeply unpopular. According to Pew, sadly only 17% of Americans think AI will have a positive impact. In China, 83% believe AI will be positive. A token tax & political backlash is coming unless the narrative changes. 🇺🇸👀🧐

@bfeld i love that you are doing this, great reading

My /start command is a 1,400-line markdown state machine with 15 steps. I mapped the data dependencies and found 12-22 seconds of overhead from defensive machinery that no longer needed to exist. adventuresinclaude.ai/posts/2026-03-…

@soumeya @altcap Interesting framing. What is it about AI that maps to this? Or put differently, what is true about AI that wasn't also true about computers as they entered the economy? Genuinely asking.

Treating this as a 'narrative' problem is exactly why the 17% exists. It frames a structural failure as a PR hurdle. The public doesn't have a story problem; they have an alignment problem. We have built a product and policy framework designed for the extraction of value by the few, while social and economic disruption is offloaded onto the millions. You don't fix that with a better ad campaign; you fix it with a better value distribution. The 83% in China exists because their narrative is tied to national strength and collective progress.

For the longest time the ideal size of an engineering pod is 6-8. What is it now? Two?

@dadgumboxscores The very odd thing for this particular game is the beginning of the second half. This coaching staff, and this team, has shown elite ability to adjust to the game and come out swinging in the second half. That just didn't happen this time. Very unsure why.

Clemson 80, Carolina 79 possession plot Carolina scored 19 points the final 15 minutes of the 1st half, and 25 points in the final 5 minutes of game time End of the 1st half was some of the worst consecutive possessions of basketball the team has played all season

@amasad 👏 just, keep, shipping. :)

We’ve raised $400M at a $9B valuation. Investors include Georgian, G Squared, Prysm, 1789, YC, Coatue, a16z, Craft, and QIA, with strategic investments from Accenture, Databricks, Okta, and Tether. We’re also lucky to have incredible individuals backing us, including Shaq and Jared Leto. This funding will help us scale our ambition and expand beyond coding into AI systems that center human creativity. Replit is now used at 85% of the Fortune 500. We have an opportunity to help shape the future of work. One where AI abstracts away the boring parts and humans shine as creative directors. We’re also investing more globally, particularly in Europe, Asia, and the Middle East. Innovation can come from anywhere in the world, and we want to help unlock it.