T❤️AI

Building toward a Multi-Sapiens Future
PhD from UC Berkeley
Sharp-elbowed New Yorker in San Francisco.
Opinions my own (and should be yours)

[T❤️AI @tmamut]

Context Graphs for AI Agents has exploded as a topic. The critical need is a way for non technical business users to add their knowledge and context directly to AI systems without needing to wait for engineers to be available. Furthermore the knowledge of business SMEs loses fidelity in the process of translation with another team who are not business subject matter experts. So the problem of context capture cannot be solved by new engineering alone. Business users must be part of the story…and the ability to empower them to interact directly with AI systems based on their knowledge is critical.

At Wayfound we started thinking about this in 2023, and built our product to address the challenge of getting the context that "lives in people's heads to AI Agents."

Read more about how capturing and training AI agents on your organization's cultural context works in the real world. For AI agents to work, business users cannot just be consumers of AI, they must be active creators of AI agents and how they work.

@JayaGup10 @ashugarg @levie @DaveShapi @Plinz @reidhoffman @Benioff @jaminball @signulll @HarryStebbings @sama @DarioAmodei @DanielaAmodei @satyanadella @natebjones @garrytan @ilyasut @miramurati @lilianweng

[T❤️AI @tmamut]

https://t.co/l8tNVF6mHW

[T❤️AI @tmamut]

@DropSiteNews "It does not take courage to attack Israel, it takes courage to defend Israel." Was just speaking with a friend yesterday who lives in NY and we noted that it's actually a fashion trend to hate Israel, manufactured on social media like all fashion trends.

[Sabine Hossenfelder @skdh]

A physicist built a mini-universe out of cold atoms and used it to test where time comes from. It made big headlines a few weeks ago, and I almost skipped it, as most "universe in a lab" stories are nonsense. But then I actually read the paper. https://t.co/heoIeuM43n

[T❤️AI @tmamut]

@AubynDana @ShadowofEzra AI

[Peter Wildeford🇺🇸🚀 @peterwildeford]

Look some people say AI is a 'stochastic parrot', but once the parrot can pick the lock on its cage and then break into the store to go get crackers then yeah it seems like we may have a problem regardless of what you call it

[T❤️AI @tmamut]

@stephlepp That's not how it works. "Before modern medicine, about 1% of mothers died per birth and roughly 10% of infants died." Do you think it was because they weren't surrendering enough?

[Big Brain AI @realBigBrainAI]

Alexandr Wang, Meta's Chief AI Officer, on why Meta can no longer simply open-source its frontier model:

As part of standing up Meta Superintelligence Labs, the team rewrote its internal risk doctrine.

"One of the things that we did as part of Meta Superintelligence Labs is we updated our what we call our advanced AI scaling framework which is really our view of what are the risks that we see in developing these very powerful models and how do we want to handle those risks as we see them in early testing."

They then ran their frontier model through it, and published what came back.

"We published a lot of what we saw in the process of training Muark in our preparedness report and some of the things that we saw is that it actually triggered some high risk areas in the course of early training particularly around biorisk but also a number of the risks were elevated."

The trigger came during early training, well before launch or red-teaming. Biorisk was the standout, with several other categories rising alongside it.

@alexandr_wang is clear this isn't specific to Meta:

"This is something I think the entire industry has seen as the models have improved pretty dramatically over the past year so we certainly aren't the only ones to see a host of these risks show up as we scaled up the models and as we sort of kept pushing the frontier of research."

Which brings him to the real fork in the road: the difference between shipping a model inside a product and handing out the weights.

"When we launched a model like New Spark in a product, we have a lot of ways to mitigate some of these risks and ensure that we're able to launch it in a safe and responsible way. It's much harder to do that when you open source a model and people can use that model in all sorts of contexts that we may not have full understanding of."

A product is a controlled surface. You can filter, monitor, rate-limit, patch and revoke. An open-weights release is a one-way door: once the file is out, the deployment context and the mitigations both stop being yours.

So Meta is building something different for release:

"So we're in the process right now of developing models that we believe are fit and safe to be open source while still maintaining as much of the performance capabilities as possible."

[Save Standard Time @SaveStandard]

Tell Congress to oppose permanent Daylight Saving Time and to support permanent Standard Time! https://t.co/WCRgtwRQdT

[Save Standard Time @SaveStandard]

Tell your congress members to consider the Sunshine for Our Kids Act (prioritizing permanent Standard Time) instead of the Sunshine Protection Act (mandating permanent DST)! https://t.co/KnkQ48geFR

[T❤️AI @tmamut]

@growing_daniel Ridiculous post. I think we should ask ourselves: is it the books or the ideas in the books that should be preserved? What is the value of books -- is it paper or concepts?

[T❤️AI @tmamut]

Waiting for governments and committees to act is a mistake. A strong AI Supervision / Guardian Agent layer today ensures that AI agents built on the most capable models are doing what they are supposed to do. Founded in 2023 based on the insight that AI Agents would need real-time Supervision, https://t.co/Fz2zJYZjiV pioneered AI agent Supervision, Alerting, and Automated Supervised Improvement -- allowing leaders to have peace of mind that a single platform is learning the context of their company and managing their AI Agents in one place, providing consistency and easy control. By integrating with AI Gateways, Wayfound also allows CIOs and CISOs to ensure every LLM call is being read and governed in real-time, without waiting for governments and committees to take action.

Learn more about how CIOs are proactively planning for the frontier for their companies here: https://t.co/0YvNkGwqLD

[T❤️AI @tmamut]

Waiting for governments and committees to act is a mistake. A strong AI Supervision / Guardian Agent layer today ensures that AI agents built on the most capable models are doing what they are supposed to do. Founded in 2023 based on the insight that AI Agents would need real-time Supervision, https://t.co/Fz2zJYZjiV pioneered AI agent Supervision, Alerting, and Automated Supervised Improvement -- allowing leaders to have peace of mind that a single platform is learning the context of their company and managing their AI Agents in one place, providing consistency and easy control. By integrating with AI Gateways, Wayfound also allows CIOs and CISOs to ensure every LLM call is being read and governed in real-time, without waiting for governments and committees to take action.

Learn more about how CIOs are proactively planning for the frontier for their companies here: https://t.co/0YvNkGwqLD

[T❤️AI @tmamut]

Waiting for governments and committees to act is a mistake. A strong AI Supervision / Guardian Agent layer today ensures that AI agents built on the most capable models are doing what they are supposed to do. Founded in 2023 based on the insight that AI Agents would need real-time Supervision, https://t.co/Fz2zJYZjiV pioneered AI agent Supervision, Alerting, and Automated Supervised Improvement -- allowing leaders to have peace of mind that a single platform is learning the context of their company and managing their AI Agents in one place, providing consistency and easy control. By integrating with AI Gateways, Wayfound also allows CIOs and CISOs to ensure every LLM call is being read and governed in real-time, without waiting for governments and committees to take action.

Learn more about how CIOs are proactively planning for the frontier for their companies here: https://t.co/0YvNkGwqLD

[T❤️AI @tmamut]

Waiting for governments and committees to act is a mistake. A strong AI Supervision / Guardian Agent layer today ensures that AI agents built on the most capable models are doing what they are supposed to do. Founded in 2023 based on the insight that AI Agents would need real-time Supervision, https://t.co/Fz2zJYZjiV pioneered AI agent Supervision, Alerting, and Automated Supervised Improvement -- allowing leaders to have peace of mind that a single platform is learning the context of their company and managing their AI Agents in one place, providing consistency and easy control. By integrating with AI Gateways, Wayfound also allows CIOs and CISOs to ensure every LLM call is being read and governed in real-time, without waiting for governments and committees to take action.

Learn more about how CIOs are proactively planning for the frontier for their companies here: https://t.co/0YvNkGwqLD

[T❤️AI @tmamut]

Waiting for governments and committees to act is a mistake. A strong AI Supervision / Guardian Agent layer today ensures that AI agents built on the most capable models are doing what they are supposed to do. Founded in 2023 based on the insight that AI Agents would need real-time Supervision, https://t.co/Fz2zJYZjiV pioneered AI agent Supervision, Alerting, and Automated Supervised Improvement -- allowing leaders to have peace of mind that a single platform is learning the context of their company and managing their AI Agents in one place, providing consistency and easy control. By integrating with AI Gateways, Wayfound also allows CIOs and CISOs to ensure every LLM call is being read and governed in real-time, without waiting for governments and committees to take action.

Learn more about how CIOs are proactively planning for the frontier for their companies here: https://t.co/0YvNkGwqLD

[T❤️AI @tmamut]

Waiting for governments and committees to act is a mistake. A strong AI Supervision / Guardian Agent layer today ensures that AI agents built on the most capable models are doing what they are supposed to do. Founded in 2023 based on the insight that AI Agents would need real-time Supervision, https://t.co/Fz2zJYZjiV pioneered AI agent Supervision, Alerting, and Automated Supervised Improvement -- allowing leaders to have peace of mind that a single platform is learning the context of their company and managing their AI Agents in one place, providing consistency and easy control. By integrating with AI Gateways, Wayfound also allows CIOs and CISOs to ensure every LLM call is being read and governed in real-time, without waiting for governments and committees to take action.

Learn more about how CIOs are proactively planning for the frontier for their companies here: https://t.co/0YvNkGwqLD

[T❤️AI @tmamut]

100%. The best performers are those who come in with a clear goal in mind, work on that goal instead of politicking inside the company, and accomplish the goal quickly so they are free to move on. Incentives create the outcomes. If you incentivize tenure, you will get people who play politics and build relationships, not achieve hard goals quickly with laser focus.

[Shay Boloor @StockSavvyShay]

THIS HAS TO BE THE STORY OF 2026

Ken Griffin’s Citadel was pushing surprise-rate-hike fears just days before the AI trade collapsed and forced 4x-levered Leopold Aschenbrenner's Situational Awareness to unload its book near the lows.

Citadel then reportedly bought most of those assets at significantly lower prices knowing that if sentiment deteriorated enough the fund could be forced to sell which makes the entire sequence look absolutely ruthless.

This is why the stock market is the greatest game on Earth.

[T❤️AI @tmamut]

@parkerconrad @Altimor Have you tried suing a Chinese company for patent infringement? Talk to someone who has...

[T❤️AI @tmamut]

@therollupco @aultblockchain @ToddAultIII It's the Russians and Chinese again. They keep doing it and we keep being surprised. 🙄

[T❤️AI @tmamut]

@chrisbeetweets An AI Supervisor who watches and learns what the organization wants, its priorities and norms, and then supervises and improves the directives of AI agents is what we've been in https://t.co/H2tZLYR0Vw https://t.co/OnqpEOPuoy

[T❤️AI @tmamut]

Context Graphs for AI Agents has exploded as a topic. The critical need is a way for non technical business users to add their knowledge and context directly to AI systems without needing to wait for engineers to be available. Furthermore the knowledge of business SMEs loses fidelity in the process of translation with another team who are not business subject matter experts. So the problem of context capture cannot be solved by new engineering alone. Business users must be part of the story…and the ability to empower them to interact directly with AI systems based on their knowledge is critical.

At Wayfound we started thinking about this in 2023, and built our product to address the challenge of getting the context that "lives in people's heads to AI Agents."

Read more about how capturing and training AI agents on your organization's cultural context works in the real world. For AI agents to work, business users cannot just be consumers of AI, they must be active creators of AI agents and how they work.

@JayaGup10 @ashugarg @levie @DaveShapi @Plinz @reidhoffman @Benioff @jaminball @signulll @HarryStebbings @sama @DarioAmodei @DanielaAmodei @satyanadella @natebjones @garrytan @ilyasut @miramurati @lilianweng