Sabitlenmiş Tweet
we raised a $13m seed round from 120+ of Silicon Valley’s top investors
for @mastra, the leading TypeScript agent framework
English
Sam Bhagwat
13.4K posts
Sam Bhagwat
@calcsam
building @mastra. author principles of building ai agents, the "most popular book in SF right now". prev cofounder @gatsbyjs
Katılım Ağustos 2012
2.6K Takip Edilen13.8K Takipçiler
@iamjakestream In our batch (W25) most startups just priced at what their partner recommended
English
This is downstream of the YC uncapped note terms. Founders rationally push for highest price possible after DD, so that the YC MFN converts for the least dilution possible.
However sets the bar *very high* for subsequent funding and results in a lot of seed extension attempts
nikhil@uninsightful
the default yc round this batch (W26) seems like 4m on 40m I remember when I first started in venture exactly three years ago (W23 batch) and most venture ppl were complaining about YC pushing their founders to do 2m on 20m in 3 years the market went from a very begrudging 2 on 20 to a more neutral 4 on 40 interesting to think about where things land 3 years from here
English
English
As always, credit to @dexhorthy for the 'Dumb Zone' phrase
English
Doing some experiments today with Opus 4.6's 1M context window.
Trying to push coding sessions deep into what I would consider the 'dumb zone' of SOTA models: >100K tokens.
The drop-off in quality is really noticeable. Dumber decisions, worse code, worse instruction-following.
Don't treat 1M context window any differently.
It's still 100K of smart, and 900K of dumb.
English
Sam Bhagwat retweetledi
Sam Bhagwat retweetledi
We've made it much easier to self-deploy Studio and add authentication 🔒
You can now securely share access to your agents with your team during development
Invite product and data people with read-only access to inspect traces and datasets without things getting messy
English
Sam Bhagwat retweetledi
Two new additions to my library by @calcsam, if anything these will be cool to look back at in 2/5/10 years to compare how far we've come and how the practices of building agents changed: RAG, MCP, orchestration...
English
@SergeyNovikov @jslishi @thekitze we have folks using this harness API and then some use the createMastraCode to embed a headless Mastra code in their app ( this is what the Superset folks do for example)
English
@jslishi @thekitze @calcsam Interesting! In fact, mastra.ai/reference/harn… Alpha, but whatever. Would be useful for many things.
But still, to have real dev orchestration - there is a lot of work to do on top of it. Unless someone already did it.
English
vibe coders who don’t ship anything showing their agent orchestration setup
English
Sam Bhagwat retweetledi
Stoked to announce remote sandboxes 🔥
Give your agent a disposable computer to safely run code and shell commands on untrusted input
Each sandbox has an isolated filesystem - or mount S3/GCS to seed files and skills, share state across sandboxes, and persist data between runs
English
we’ve partnered with @1Password to take the next step toward solving authentication for agents.
last month, we introduced managed auth: a standardized way for agents to log in and stay logged in across the internet.
with this partnership, your agents can now use credentials directly from your 1password vaults with managed auth.
English
The technical depth of @darian314 and team shows in the companies they invest in.
Some of the highest caliber founders I've met have been in the @gradientvc family.
Glad to have them in Mastra!
Darian Shirazi@darian314
We're thrilled to announce @GradientVC's Fund 5: $220M, oversubscribed, and fully independent. As @agarfinks wrote in @FortuneMagazine's Term Sheet this morning — we founded Gradient when nobody took AI seriously. Turns out that was the point. 📰 fortune.com/2026/03/17/goo…
English
Even bigger irony of getting rich is that everything expensive isn't that much better than when you paid normal for it
Many things are even worse (most expensive luxury hotels are guaranteed worse than regular simple hotels, I know I tried most of them now)
The real reason you wanna get rich is not to buy expensive things
It's so that $1M invested gives you 3% to take out every year with no risk, which is $30,000/year
Which you can use to travel for $1000/mo on a shoestring budget forever without having to back to some desk job with a shitty boss
Aka FREEDOM
@levelsio@levelsio
The irony is that traveling on <$1000/mo is way more fun than >$10,000/mo Luxury travel is extremely boring, comfortable, not challenging, sycophantic (yes sir) Travel on a shoestring budget you get inventive, are forced to meet locals just to survive and get around, have to hitchhike etc I like to combine cheap and luxury travel which keeps my brain from decaying and the contrast actually lets you enjoy both
English
@calcsam @LAUNCH @askgrapple thanks sam! couldn’t have done it without the support of the three amigos @smthomas3 @abhiaiyer
English
A couple months ago @LAUNCH invested in @askgrapple, and I’m in SF this week to kick-off LAUNCH accelerator (LA36)!
At Grapple, we’re building the most vertically integrated data analytics product and automating it with AI so sales, marketing, and finance leaders can get into the weeds and understand their data.
Sounds easy but the data industrial complex wants to sell you over a dozen different tools just for basic analysis. Companies spend millions on a data warehouse, pipeline, modeling, orchestration, BI just to have rigid out-of-date dashboards.
English
Probably going live in an hour. What should I talk about today?
English
Similarity isn’t the same as understanding — real context comes from relationships.
Excited about what Nishkarsh and the HydraDB team are building and proud to be an investor!
Nishkarsh@contextkingceo
We've raised $6.5M to kill vector databases. Every system today retrieves context the same way: vector search that stores everything as flat embeddings and returns whatever "feels" closest. Similar, sure. Relevant? Almost never. Embeddings can’t tell a Q3 renewal clause from a Q1 termination notice if the language is close enough. A friend of mine asked his AI about a contract last week, and it returned a detailed, perfectly crafted answer pulled from a completely different client’s file. Once you’re dealing with 10M+ documents, these mix-ups happen all the time. VectorDB accuracy goes to shit. We built @hydra_db for exactly this. HydraDB builds an ontology-first context graph over your data, maps relationships between entities, understands the 'why' behind documents, and tracks how information evolves over time. So when you ask about 'Apple,' it knows you mean the company you're serving as a customer. Not the fruit. Even when a vector DB's similarity score says 0.94. More below ⬇️
English