Thank you to all of the wonderful founders and friends of FirstMark who joined us at Dante last night for the 2026 Founder Dinner. It was a fantastic night for all of the FirstMark community. See you next year! đź—˝
FirstMark CEO Summit today feat the iconic @harleyf of @Shopify, Mark Ebert of @profound, Alex Buder Shapiro of @clay, and @glencoates of OpenAI 🚀
Always an incredible day bringing the FirstMark family together under one roof!
FirstMark retweeted
Crazy launch! Copycats! Massive round! On the one-month anniversary of the launch of Jev, we unpack it all with @CompleteSkeptic, CEO of @typesafeai: what Jev is, how it fixes what's broken in AI, and why the fast-followers are missing the hard part.
00:00 “95% Einstein, 5% Mr. Bean”
1:09 Why intelligence per dollar matters
3:25 Why TypeSafe built in stealth for two years
5:50 “We could have had AGI last year”
6:48 Why the product carries the marketing
9:33 What is Jev? Machine-native AI, built for software
10:53 System One vs. System Two
12:26 From simple decisions to composable AI
15:21 Automating email: giant prompts vs. smart if-statements
20:03 Probabilities, thresholds, and developer control
23:43 Why Jev is faster and cheaper than chat models
25:39 “Classification is AI-complete”
27:16 RLHF: how OpenAI trained models to please humans
31:28 Sycophancy, refusals, and the hidden costs of RLHF
34:22 RLVR: why benchmark gains don’t mean reliable automation
37:51 RLCD: optimizing for machine-native intelligence
39:54 Why TypeSafe builds its training data in-house
41:49 The Bitterest Lesson: define the task before scaling
44:39 Building Jev from open-weight models
45:56 Is Jev just a classifier? Why that’s a compliment
48:35 “Build prod, not God”
50:11 Embedding AI in software, beyond coding assistants
53:05 Jev’s copycats: matching the interface vs. the intelligence
55:33 Why benchmarks mislead the users they should help
58:37 Open weights vs. APIs: how TypeSafe weighs the tradeoff
1:00:04 Jevons paradox: a new market for machine intelligence
1:02:50 Why Diogo is skeptical of recursive self-improvement
1:04:47 Can practical automation avert an AI winter?
FirstMark retweeted
Huge congrats to @eglyman, @karimatiyeh, and the amazing @tryramp team!
Incredibly rare to combine this level of talent density, product velocity and relentless execution, esp at this scale. Thrilled to be part of the journey 🚀
FirstMark retweeted
Huge thanks to @tbpn @jordihays and @johncoogan
for covering @LedgerAPI's funding announcement yesterday
FirstMark retweeted
We are replacing RocksDB.
Our new persistent storage engine is out today, open source under Apache 2.0.
With every commit flushed to disk: reads 7.95x faster, updates 6.10x faster, creates 26.8x faster.
Read the post: surrealdb.com/blog/we-are-re…
Speed or durability? Yes.
We're replacing RocksDB.
New persistent storage engine: reads 8.0x and updates 6.1x faster than RocksDB. Reads 73.7x faster than SlateDB (on disk, not object storage).
Run it yourself.
Stop choosing.
sdb.li/3VE7i24
FirstMark retweeted
Replying to @neckelthomas @FirstMarkCap @BoxGroup @AdamRothenberg @ClaireSmilow @gokulr @jaltma @cjc @emiliechoi @maxbranzburg @MBTannenbaum @NikMilanovic @jeff_weinstein @hongxia_zhong @getCos
and huge shoutout to @c_grigera for the awesome article in Fortune
fortune.com/2026/10/14/fragm…
FirstMark retweeted
Fragment is building the infrastructures to make agents financial actors. Excited to be leading their Series A and partnering with @neckelthomas
fortune.com/2026/10/14/fragm…
Everyone’s chatbot can book a doctor’s appointment.
The real test is whether you can earn the trust of a Spanish-speaking diabetic woman, walk her through a Medicaid application while finding her secure housing, and coach her into choosing a 12oz (instead of 16oz) soda along the way.
Today we’re sharing BabaBench, a benchmark measuring how frontier models perform the real work of healthcare coordination. Our synthetic worlds are built from annotations by Baba’s in-house team of hundreds of expert doctors, therapists, nurses, and social workers.
Turns out, AI alone isn’t ready to replace human care coordination.
🔥🔥🔥
ChatGPT for video has finally arrived after +9 years in the making:
synthesia.io/syren
Can't wait to see what people create🥳
FirstMark retweeted
Launch week starts on Monday. It is the biggest thing we have shipped as a company.
I spend most days with customers and partners, and the pattern is always the same. Teams are forced to compromise on things they should never have to compromise on.
Next week we answer that head on.
12-15 October.
surrealdb.com/stop-choosing
See you next week.
Speed or durability?
In-memory or persistent?
Simplicity or scale?
Memory or accuracy?
Transactions or speed?
SQL or SurrealQL?
Yes.
Stop choosing.
sdb.li/4eicEGr
12-16 October. See you next week.
FirstMark retweeted
The center of the current discussion within the fintech ecosystem is undeniably stablecoins and I've often said, both internally and externally, that fintech is crypto and crypto is fintech.
We should expect these two worlds to continue to collide and overlap from here on out, which means there will be incredible opportunity for new companies that bridge these worlds. Here at @FirstMarkCap we are already seeing this transition play out. There are companies, both in our portfolio and outside of it, that are using stablecoins as a new financial primitive to deliver for end users (both consumers and businesses).
As a result of this unloc, a new crop of stablecoin native and onchain native-businesses has emerged with unique financing needs and preferences. Those needs create opportunities for new businesses like @ValinorDigital, which is exciting to us because we firmly believe that the next big opportunity within fintech more broadly is onchain credit.
Valinor is a fascinating business. Their north star is to build a modern credit institution, and they have a core belief that all of capital markets will move onchain. They want to be one of the first (and best) institutionally focused providers to leverage blockchain-based rails (and stablecoins) to deliver cheaper, better, and faster credit solutions to any type of business around the world. Ive gotten to know the founders, Connor Dougherty and Lily Yarborough, over the past year and have been consistently impressed by their thoughtfulness, open-mindedness, and rigor.
I recently had the chance to chat in more depth with Connor about his entrepreneurial influences, how they think about using stablecoins for lending, who they are serving as customers, and how theyre applying AI in their business today.This was a great conversation and a peek inside the early days of a company truly at the frontier of fintech
FirstMark retweeted
Special treat for the @andy_pavlo fans out there: an in-depth discussion about what happens to databases when billions of AI agents hit them, his move to build a research lab at @ClickHouseDB, and plently of hot takes on the database market (Postgres, vector, graph, GPU etc)
00:00 Cold open & Intro
01:21 From vector databases and RAG to the age of agents
04:09 Neon's stat: agents create 80% of databases? 07:25 Why agents keep deleting production databases 08:19 Guardrails: the toddler-and-stairs rule
10:51 The four eras of database volume: 10–100x more queries
15:10 Agent memory: files vs. databases ("everything is a database")
18:24 Which database do AI models recommend?
19:25 "It cites me back to myself"
23:14 MCP for databases, and text-to-SQL from 60% to 99.5%
27:09 Trust an agent the way you'd trust a junior developer
27:47 Can AI build an entire database? Opus 4 and the CMU projects
29:57 60% of open-source databases now have AI commits
30:43 Ten years of self-driving databases, Peloton to today
33:52 What LLMs changed: 85% of the tuning in 15 minutes
35:42 Should anyone still study databases?
38:27 Free CMU courses, the DJ, and the Wu-Tang final exam
42:09 Why start a research lab inside ClickHouse?
45:53 Why ClickHouse looked like vaporware in 2016
47:46 What makes ClickHouse fast: columns, vectors, Snowflake's lineage
52:22 Why Databricks, Snowflake and ClickHouse all added Postgres
58:15 Vector, graph and GPU databases: thumbs up or down?
1:08:34 Is the database market stagnant? "A cheetah on cocaine in a Ferrari"
1:10:14 The relational model is arithmetic; SQL as the new assembly
1:11:59 Larry Ellison, Linux, and why databases still matter
FirstMark retweeted
The way we're currently growing, we'd hit a triple-digit run rate by the end of the year. And 5x that by April '27 if we stay on course.
I have no idea how the math keeps working out. But the dashboard doesn't lie.
For the last few months, I've looked at the numbers, thinking there's no way this is real. But we kept hitting the projections.
And to think we pivoted from a product at ~$300k ARR last year makes it feel even more unreal.
Easily the most fun I've had in life.
FirstMark retweeted
News: I've founded a new startup and raised pre-seed funding led by @FirstMarkCap. Meet @piccardscience – Physical AI for ocean robotics.
I grew up mesmerized by Jacques Cousteau's underwater explorations. I regularly fall asleep listening to Sir David Attenborough's soothing voice narrating Blue Planet.
My wife and I lived in 13 countries by the ocean, surfing and scuba diving. We named our son Kai, which means ocean in Hawaiian.
We left our life in New York and moved to Rhode Island, The Ocean State, to raise Kai by the ocean in a small coastal town with a surf culture instead of in the city.
And like magic, our personal life choices all of a sudden intersected with my professional path.
Over the summer I got invited to teach agentic AI to professors at University of Rhode Island. While preparing, I found out URI was the first institution in the US to establish M.S. and PhD degrees in ocean engineering and today they have one of the top programs in the field.
In June, they cut the ribbon on a brand-new ocean robotics lab building as part of a $300M plan for the revitalization of the Narragansett Bay Campus, down the road from our house.
The lab is home to research projects I can only describe as...
Cool AF.
After starting to work with the researchers at the lab, it became clear what needed to happen.
The ocean is like space. Full of unknowns and mysteries. Incredible opportunity for exploration. The new advances in AI make it possible to do things that were previously impossible, too expensive or too dangerous.
I believe in a big future for autonomous oceans.
And so Piccard was born. We'll first focus on the the physical AI involved in building ocean robotics, such as autonomous underwater and surface vehicles. Piccard has big aspirations which I'll reveal in time.
When I told my long-time investor @amishjani at FirstMark about it, he immediately saw the potential and led a round that came together in days.
Along with long-time supporters who've believed in me and invested in my startups for years. Dustin Dolginow and @anabelpaksoy of 43, @hnada, @mutaharglasgow, @paulsethi, and @smart .
So now we start this journey.
Why the name? Jacques Piccard, along with U.S. Navy's Don Walsh, was the first person to reach the deepest point in the ocean, the Challenger Deep in the Mariana Trench, in 1960. He is an inspiration.
Piccard also got accepted into NVIDIA's Inception program. We'll be building across the @NVIDIARobotics physical AI stack, and you'll see me share a lot of content about this.
Ocean robotics is difficult science. I've become a student again. Taking classes at URI in Underwater Acoustics and Oceanography Data too.
If you're intrigued by physical AI and want to understand why Jensen Huang always talks about it, turn on notifications for my posts. I'm sure you'll learn a ton through them.
If you're an ocean robotics researcher or a builder of autonomous ocean vehicles, DM me. You'll like what I have for you.
FirstMark retweeted
Ofc everything is on X is mind-blowing, but having seen the beta version, this is... truly mind-blowing.
Holy S… we cracked on-brand AI video 🫨
$20K agency videos for $5 in 5 mins.
Introducing Synthesia Syren: one prompt → fully on-brand video.
Upload existing videos as Brand Context.
Syren learns motion, pace, tone of voice, colors and fonts.
Launching Friday: Join the waitlist now to get your $50 in free credits: synthesia.io/syren
FirstMark retweeted
Our latest TTS model has been out for two weeks, and it still leads the @voicearena_ai US English latency board.
Sub-100ms median time to first audio, down from 236ms for our previous model.
Try it: gradium.link/studio
Gradium-TTS is #1 on the Voice Arena US English 🇺🇸 latency board: 68 ms to first audio at the median, the fastest model we have measured.
Gradium-TTS is the latest TTS model from @GradiumAI. It comes in ahead of SpeechifyAI's Simba 3.2 at 123 ms and Inworld TTS 2 at 168.5 ms, and more than three times faster than its own earlier 236 ms.
Speed without giving up quality: on the same board it ranks #10 of 21 at 1007 Elo from blind native-speaker votes, which puts it on the quality-versus-speed frontier with Simba 3.2, Inworld TTS 2 and Cartesia's Sonic-3.6.
We measure time to first audio the way a listener hears it, with leading silence counted and fresh text on every request so nothing is served from cache.
Congratulations to @GradiumAI on the release!
See below for the quality board, what raters said, and the exact audio they judged đź§µ
FirstMark retweeted
Almost every AI company or inference cloud we talk to has the same problem: they raise a round, sign a handful of large compute contracts, and a big chunk of the money goes straight into a deposit before a single GPU is turned on. Equity is the most expensive capital a startup has, and it's being spent on a bill that comes due in months. No other commodity has been so reliant on a single source of funding.
Today that changes. Alongside K8, we've launched a $250M prepayment facility that lets AI companies and inference clouds finance their compute deposits against the contract instead of paying them out of their round. The operator still gets paid upfront. The buyer keeps its equity for research, hiring and product.
It's the first financial product we've built on top of our physical exchange, and it won't be the last. Thank you to the K8 team for underwriting something that hadn't been done before.
Liquid Compute and K8 announce a $250M prepayment facility to help compute buyers fund prepayment capacity commitments.
This facility makes compute financeable the way every other commodity already is, and we are already finalizing our first deployments with a few large inference clouds.
We're proud to build this with K8 as we bring the discipline of commodity finance to compute.




































