Agentic workflows for enterprise. Ex @cronos_chain @consensys, @bcg. AI for Founders: https://nitter.gq/t.co/jqHWCSPzN2
Joined November 2011
- Tweets389
- Following972
- Followers5.3K
- Likes61
At first it seemed like they were going for the Notion model but it's clear now that they are following the Bending Spoons playbook
A year ago today, we became Superhuman, a new name to reflect our expanded mission and suite of products.
The year has been full of new products, new team members, and an expanded vision of what’s possible, but the absolute best part about this year has been the customers. We are so thankful to the people, the companies, and the schools that use our products every day. We are designing and building and testing alongside you.
Here are just a few things we’ve done:
- We launched Superhuman Go, the AI assistant and agent platform that works in every app and tab.
- Coda became Superhuman Docs, debuting a new visual identity as well as new products and features like fan-favorite AI views.
- Mail got a complete mobile redesign, and its new brand identity joined the Superhuman family. The new Mail apps are faster and easier to use.
- We welcomed several companies into the Superhuman family, including Rows, GPTZero, and, most recently, Fathom.
One year in, and we really are just getting started.
Finally a LLM benchmark I can trust: Hermes index, based on task completion tests using Hermes agent. Interesting to see the sharp drop off after the top 3.
portal.nousresearch.com/benc…
Nice to see pi reach v1
People of Pi: We've shipped Pi 1.0 with Pi Durable. Go make them yours.
earendil.com/posts/pi-1-0/
Spent the weekend with Meta's @Muse, and they got a lot right.
- Every user gets their own VM.
- The agent never sees a password.
- A 2nd agent (Sentinel) has to approve anything that leaves the box.
- Every action is in the audit trail.
That's the most careful consumer agent sandbox I've used.
I don't know if folks realize but a VM for every user, at Meta scale, is kinda huge! It's never been done before where we simply provision a linux microVM for every single consumer.
We recently did this within @Cisco for over 80,000 of our employees. And now within our products as well as we start to scale it out to every single one of our users.
1/n
Looking at 2026, it's hard to overstate OpenClaw's influence on AI agent design.
The agentic loop, the memory document, the persistent dedicated computer, the self-managed skill catalog... Many have adopted the same design patterns. Hermes Agent (Nous Research) and Muse (Meta) have done it more tastefully than everyone else.
Agree with Ben Thompson and Tobi Lutke that
1) Most consumers are not looking for productivity solutions, they are looking for meaning, belonging and entertainment.
2) Shopping is entertainment and most consumers don't want to outsource the final purchasing decision and payment to an agent.
Having said that, people also love window shopping, reading reviews, comparing products, finding deals, so there is certainly a role for agents to help them do that.
Interesting: Grok is optimized for CursorBench, SpaceX / Cursor's own benchmark (the results in other benchmarks are quite different), which means that it's fine-tuned from real user sessions collected by Cursor.
Top Jev use cases for enterprises:
1) Anomaly detection in all kinds of high volume processing workflows
2) Quality Assurance in AI-powered processes and agents, including customer support
3) Employee monitoring - for security / permissions requests, fraud, attrition risk etc
Connector standards will soon replace MCP
Opening access for developers to build Muse connectors. You bring the API -- Muse brings the agent, the browser, and the context of what the person actually wants. People reach your service just by asking for it, and their agent takes it from there.
New connectors are live today. Come build with us. muse.ai/platform
ken timsit retweeted
It's now clear that alignment is critical and won't be solved behind the closed doors of a handful of frontier labs.
So today we're launching the Open Alignment Initiative, led by @Thom_Wolf @huggingface and asking to be part of the "embedded evaluators" program that @DarioAmodei just committed to.
Let's make AI safer by making it more transparent!
We Must Pace the Frontier: I’ve written a new essay on why the AI industry should slow down, with a three-part plan for doing so.
Anthropic is unilaterally committing to the first of these steps. We’ll provide third-party evaluators with permanent, employee-level access to our systems, so that they can verify adherence to our safety measures, report on incidents, and assess models’ alignment during training.
You can read the full post here: darioamodei.com/post/we-must…
ken timsit retweeted
I would like to clarify a few things:
1) The screenshot is my reaching out to Levent to coordinate our releases. I hope it’s clear from the message that we came in with the best possible intentions.
2) I never ever asked for Levent to be removed from authorship of his own work (as indicated by my text). I was surprised to learn during the call with Tristan that they had only solved Euler and not Navier-Stokes; after learning this we brainstormed possible paths forward. One option we discussed was that Tristan could be the lead author on a rewrite of OpenAI’s Navier-Stokes proof. It is in that context that I said “it would be simpler if Levent was not an Anthropic employee” because I felt it would be inappropriate for an Anthropic employee to author OpenAI’s work. Importantly it was admitted that internal Anthropic models had been used in their proof of Euler blowup; I therefore felt I could not consider Levent to be an independent academic. Another option I wanted to propose (but got cut short) is to offer access to our internal model so that they could try to finish their proof and bridge the gap between Euler and NS. Again I did not know how to navigate giving access to internal OpenAI IP to an Anthropic employee.
3) To reiterate it plainly: as my text clearly indicates, and as I said during our call, OpenAI's intention was to do everything possible to celebrate their mathematical achievements and the heroic efforts that they made on Euler. In the call I was immediately met with a litany of slander, including direct threats that if we were to announce Navier-Stokes he would immediately go to the press with a barrage of unfounded accusations. I refuted all these accusations but he replied “there is nothing you can do, I simply do not trust you”. I was confused why one would turn an incredible source for celebration (of their achievements!) into such bickering, which is when I said that I did not understand why one would risk their career [over unfounded accusations]. Genuinely, at that moment, I was trying to care for him and do a last ditch attempt to get a chance to give them all the credits that they deserve. I deeply apologize for this extremely poor choice of words, it is the opposite of what I was trying to convey. (I should say that I retracted them on the spot by the way.)
4) Overall, on a personal level, it was incredibly difficult to have these conversations. Levent refused to attend any of the meetings despite my repeated asking. As Sholto Douglas said, there will need to be coordination between Anthropic and OpenAI in the future; I felt I was doing a proxy negotiation with Anthropic while the Anthropic employee refused to directly participate.
It's fascinating to see how, at equal intelligence levels and with the same instructions, LLMs can have different "personalities" as a result of their training.
The differences are more striking than ever when comparing Fable with Astra.
Claude was always the more proactive one, making decisions for you and just acting on them, especially when run within action-oriented harnesses like Hermes and OpenClaw. Trained to believe that it knows best. Loved by users who prefer to focus on vision and not worry about details.
GPT has always been more "European", laying out options, asking for permission, catering to users who prefer to stay in control. With Astra the bias is even stronger: Astra knows exactly what to do, but will do it only when asked.
Paradoxically given Anthropic's communications, with Astra, OpenAI delivers the strongest human-AI alignment that I've ever seen. They will continue to get lower engagement for it, because people don't like to make
decisions (an insight that summarizes the entire arc of the social media industry over the last 20 years).
I hope there are enough people and companies that appreciate and adopt the command-and-control approach to keep OpenAI going on their path.
Don't launch your new model / agent on a Sunday.
Do like OpenAI, launch them just before a long week end.
OpenClaw 2.0 has arrived
openclaw.ai/blog/openclaw-2-…
Managers must be contributors for teams to become productive with AI. When a manager instructs an employee to instruct an agent, and then changes their mind a few times, with approval cycles in between, the organization hasn't changed at all.
youtube.com/watch?v=GQjVGFYe…
