Cofounder @getknap, @nxtblln, @payjoy, @electrobugtech, @droptalk. Formerly BD @stellarorg, @skycatch, @facebook, @google. Views are my own. markheynen.eth
7.878149,98.413454
Joined March 2009
- Tweets1.4K
- Following738
- Followers1.9K
- Likes2.1K
A month ago I argued AI needs its own trilemma, the way crypto had one: Simplicity, Safety, Sovereignty. Pick-two is the default, and pick-two is why enterprise AI stalls.
Then @grok Bot (@bot) and @jack's Berd launched days apart and drew the map for me.
@bot takes Simplicity. Agents with their own cloud computer that sign into the tools you already use. An early tester: "there wasn't anything to learn."
That is a real accomplishment and I don't want to undersell it. Zero-friction adoption is hard.
But it's a rental. 👇
Berd takes Sovereignty. @blocks open-sourced it Wednesday. Apache 2.0. Runs local. Telemetry off by default. Bring your own models and harnesses.
Honestly? This is my thesis shipping from a public company, and I'm glad it exists.
But it's not for normies. A GitHub release and a provider config as the on-ramp. A UI that deliberately surfaces failures rather than absorbing them. That's a great choice for engineers and a non-starter for the 99% who just want their email triaged.
And neither one takes Safety. Not "is the model aligned" — I mean: what stops it deleting, sending, or running the wrong thing?
That's the leg that kills deployments. It's also the least glamorous, which is exactly why it's still open.
Simplicity. Safety. Sovereignty. Two out of three does not suffice.
$10B for @OpenRouter says routing is strategic.
Worth remembering what a router is: a machine that sees every prompt you send and every answer back. Today you trust that party. After an acquisition you trust a different one, with different incentives. You don't get a vote.
Which is why @getknap builds on TrustedRouter. @jperla's design runs in an attested enclave, open source, cryptographically verifiable, fails closed. You're not trusting a privacy policy — you're checking a signature.
Routing is strategic. Verifiable routing is the version that survives.
Three years ago Daniel Lopez from @Sevanta asked me a question nobody was really asking yet: could you put a real AI assistant inside dealflow software without the data ever leaving the firm?
Sevanta Raven is the answer, and it's live. Built on @getknap.
The interesting part isn't the chat. Raven takes action across the other tools a firm actually runs on, and the underlying data never moves.
Zero data passed to models, zero data retention, zero complex setup required.
The strictest-compliance industries are where embedded agents pay off fastest. The manual work is expensive, and "just paste it into ChatGPT" or "just hook up the MCPs" isn't available.
Three good years, Daniel. Just getting going.
Fascinating to see how @sierraofficial evolved their approach to one super agent with tight privacy controls.
Also very reassuring as this is exactly what the @getknap desktop app enables via Scout, right down to the overnight dreaming.
Get in touch if interested.
We wrote up how we've AI-pilled Sierra. The centerpiece is an AI agent named Pinecone that the whole company uses for everything from business analysis to writing code. Pinecone now generates 70% of our PRs and automates hundreds of tasks every day to quietly handle work no one explicitly prompted. sierra.ai/blog/ai-pilling-ou…
Trying to manage token spend and agent use at your company?
@jperla and I are looking for 20–30 minutes of feedback from teams deploying LLMs at work.
We want to understand:
• What you need from an enterprise AI router
• Where today’s routing/gateway tools fall short
• What would make AI production-ready for your org
We’ll give you $25 in TrustedRouter API credits as a thank you.
No sales pitch—just product research.
DM me or reply if you’re interested.
As someone who was present at the creation, strongly endorse.
MoneyGram 🤝 @StellarOrg
We’re extending our five-year partnership with SDF, continuing to turn stablecoin potential into real-world utility and financial inclusion.
Building on our success in Colombia, the MoneyGram app’s stablecoin balance is now live in El Salvador.
Read more: prnewswire.com/news-releases…
Mark Heynen 🎒🦞 retweeted
It happened.
An open weights model just dropped that benchmarks higher than Opus 4.6 is out
If you have 2 Mac Studios w/ 512gb, you can run Opus 4.6 level intelligence completely for free on your desk
I warned you this would happen months ago. Now Mac Studios and Mac Minis are sold out
The next Mac Studio has been delayed until Q3/Q4. The price will be significantly higher
I told you this was going to happen. Intelligence explosion. Hardware bottleneck. Increased efficiency
Luckily I picked up 2 Mac Studio 512gbs, 2 Mac Minis, and a DGX Spark
I will be loading this up in the next couple of days and will have completely private super intelligence running for me 24/7
I’m telling you right now by end of year we will have a local version of Mythos. It’s 100% guaranteed
You called me crazy but every single prediction I’ve made has turned out to be true
These models will only get more efficient and require less hardware. But that hardware is only going to get more expensive
Local/open source is so obviously the future and if you’re still denying this now you are delusional
Meet Kimi K2.6: Advancing Open-Source Coding
🔹Open-source SOTA on HLE w/ tools (54.0), SWE-Bench Pro (58.6), SWE-bench Multilingual (76.7), BrowseComp (83.2), Toolathlon (50.0), Charxiv w/ python(86.7), Math Vision w/ python (93.2)
What's new:
🔹Long-horizon coding - 4,000+ tool calls, over 12 hours of continuous execution, with generalization across languages (Rust, Go, Python) and tasks (frontend, devops, perf optimization).
🔹Motion-rich frontend - Videos in hero sections, WebGL shaders, GSAP + Framer Motion, Three.js 3D.
🔹Agent Swarms, elevated - 300 parallel sub-agents × 4,000 steps per run (up from K2.5's 100 / 1,500). One prompt, 100+ files.
🔹Proactive Agents - K2.6 model powers OpenClaw, Hermes Agent, etc for 24/7 autonomous ops.
🔹Claw Groups (research preview) - bring your own agents, command your friends', bots & humans in the loop.
-
K2.6 is now live on kimi.com in chat mode and agent mode.
For production-grade coding, pair K2.6 with Kimi Code: kimi.com/code
-
🔗 API: platform.moonshot.ai
🔗 Tech blog: kimi.com/blog/kimi-k2-6
🔗 Weights & code: huggingface.co/moonshotai/Ki…
We are on the verge of self healing software -- you encounter a bug, it gets logged by a back end service (@sentry etc), and automatically kicks off an agent who merges in a fix. It may even happen within the user's session.
This is the age of miracle and wonder.
Mark Heynen 🎒🦞 retweeted
It’s April now
OpenClaw with docker sandbox, logging mitmproxy firewall and Clawvisor and you are good to go
The days of “it’s insecure” for OpenClaw are over
We spent a lot of time trying to solve the blockchain trilemma. It seems AI (and specifically agentic AI) has its own trilemma.
We are working hard at @getknap to solve it.
xitter.cf/getknap/status/2043753…
Mark Heynen 🎒🦞 retweeted
Episode 19: Crossing Streams
In this episode, we sit down with @markheynen and James Cantwell from the #AIforAdvisors podcast for a special crossover episode. We went straight at the questions #wealthmanagement can’t dodge anymore: the real bottlenecks slowing #AI adoption, why #vibecoding won’t magically replace #wealthtech, and what #AIagents actually look like when they’re executing across systems, plus a lot more.
🎬 YouTube: youtube.com/watch?v=ilseGZMS…
🎬 X: xitter.cf/zerobps_podcast/status…
🎧 Apple Podcasts: podcasts.apple.com/us/podcas…
🎧 Spotify: open.spotify.com/episode/1aO…
Watch the full Episode 19 of #ZeroBasisPoints!
This is a special crossover featuring @markheynen and James Cantwell from the #AIforAdvisors Podcast.
We get into the real bottlenecks slowing AI adoption, why vibe coding won’t replace wealthtech, and what agents look like when they actually execute across systems, plus a lot more.
Hosted by @haik_sahakyan and @georgeguidotti
Watch the full episode 👇
#wealthmanagement #fintech #wealthtech #AI #agenticAI #podcast
Mark Heynen 🎒🦞 retweeted
💬 “The compliance people are gonna start using these tools, and we’re gonna see a lot more adoption.”
@markheynen shares what he’s hearing firsthand from compliance teams and where the hesitation really comes from.
Watch the full discussion for the broader conversation on what’s real in AI adoption 👇
🎬 YouTube: youtu.be/ilseGZMSh2Q
🎧 Apple Podcasts: podcasts.apple.com/us/podcas…
🎧 Spotify: open.spotify.com/episode/1aO…
#wealthmanagement #fintech #wealthtech #AI #agenticAI #podcast