Adam (in SF & enjoying)

Ex-founder in AI and DeFi. Seed angel investor in @ethena & @PrimeIntellect.

Top 1% @Cursor_ai user by token usage in my city.

[Adam (in SF & enjoying) @adamdbs]

coding with extreme haste. I have 100 users in the last 1hr.

[Adam (in SF & enjoying) @adamdbs]

@cherdougie put it on a billboard in NoDi

[Adam (in SF & enjoying) @adamdbs]

In the near future I think we will have a much closer relationship with AI, cancer vaccines and genome sequencing. This guide is the tip of the iceberg. https://t.co/hKG1T4nDu9

[Susan Zhang @suchenzang]

anthropic doesn't owe anyone "frontier capabilities".

none of the labs do.

they are all simply selling a product, or a story, that people pay for.

that aside, the more telling bit is how far anthropic is willing to go to secure a narrative around "capability slowdown", post a massive raise, before an ipo, and with enterprise contracts rising for those rich enough to pay to similarly keep up the image of "powered-by/secured-by agentic AI".

with the amount of capex spent so far, this was never meant to be some democratizing technology "for the people".

this is all simply just business.

[Adam (in SF & enjoying) @adamdbs]

It feels like today is a very significant moment for closed AI models vs Open. I don't have any great takes here beyond that Anthropic is a business and will continue to try to protect its future revenue streams.

[Soumye Singhal @soumyesinghal]

This is why open-source AI matters. If the tools for building the future can be silently throttled by the same few labs competing to own that future, that’s not democratizing intelligence instead it’s building a new colonial infrastructure.

[Adam (in SF & enjoying) @adamdbs]

@finbarrtimbers Open source models getting better because of Mythos, Openrouter gaining in popularity and their revenue shrinking over a medium time horizon.

[Adam (in SF & enjoying) @adamdbs]

must avoid tyrannical researcher fiefdom at all costs

[Mark Saroufim @marksaroufim]

June 9th Researcher Reciprocity License

"if you train on it, you let us generate - reverse terms of use void"

Status quo
1. We teach frontier devs with ICLR/NeurIPS papers, OSS Github contributions
2. They use it to make frontier models
3. Then ban us from exploring our ideas

We need a new license, original thinkers can't be an underclass to a tyrannical researcher fiefdom

[Adam (in SF & enjoying) @adamdbs]

@FanFarting @gnosispay @Plasma Its okay I made the payment!

[Adam (in SF & enjoying) @adamdbs]

I had to buy a URL for $1. @gnosispay is still bricked for me until support responds. Revulot refuses to make the payment for an unclear reason.

Only option left is @Plasma One and it delivered.

[Adam (in SF & enjoying) @adamdbs]

@parisbayarea He came, he engaged 7M, he left.

[Adam (in SF & enjoying) @adamdbs]

@lilianr Such an important product for the iPhone! it is nice to see the segment get strong leadership.

[Adam (in SF & enjoying) @adamdbs]

Permissionless financial markets at scale is amazing for some use cases (hyperliquid) and catastrophic in others.

[Taylor Lorenz @TaylorLorenz]

Let's check in on how this product is going! (thread)

[Adam (in SF & enjoying) @adamdbs]

@arcee_ai @samfrannn Congrats to @samfrannn!

[Adam (in SF & enjoying) @adamdbs]

It's amazing how far OpenAI has come. This use to be the most promising work they were putting out and now we almost have AGI for coding.
https://t.co/pCyd8iVATz

[Adam (in SF & enjoying) @adamdbs]

The Anthropic deal with SpaceX highlights how crunched the market is to secure large GPU clusters. This dynamic can't last forever unless the underlying resources (energy etc. ) can't grow to meet demand.

[Adam (in SF & enjoying) @adamdbs]

What will win Brian's world:
- task-specific fine-tuned open models (RL companies that execute)
- routing to the right model for the job (OpenRouter)
- Compute providers

Open Q for me is when the supply chain can meet intelligence demand.

[Brian Armstrong @brian_armstrong]

Good take

My guess is
- demand for intelligence is near infinite
- but 80% of workloads will be running on 99% cheaper models within 12-18 months
- 20% of workloads will still run on latest gen models where IQ maxing is important (scientific breakthroughs, higher level ochestrator agents?)
- rough analogy might be what % of macbooks or gaming PCs sold have the maxed out specs for CPU/GPU, prices are falling much faster than Moore's law here though
- this leads me to think the limiting factor will be energy and compute, not better models

At Coinbase we're working hard on routing prompts to cheaper models where appropriate, and in some cases have been able to keep costs roughly flat, while token usage continues to grow exponentially.

[Adam (in SF & enjoying) @adamdbs]

I am a huge fan of Gnosis and the team there. The events with gnosis pay though highlighted point 3. In many ways this space is too early for actual adoption. There needs to be insurance and it should be abstracted from the user.

[pray.eth @pray_eth]

Thoughts on the @gnosispay hack:

1 - @gnosis_ team is honorable, they keep their word, users should not panic. However, after 5 years, so much in DeFi still relies on the integrity of the core team, which is unfortunate.

2 - Communication was excellent by crypto standards, but not great for a non-crypto audience. I followed updates on Twitter and Telegram, I understood what was likely happening behind the scenes.

Non-crypto people will need emergency numbers to call, proactive email updates, website updates, etc. They would not know, for example, that their other accounts (non crypto) are not impacted, or understand the timing involved in white hat rescue. They would not know to follow @koeppelmann on Twitter for updates, etc. It more or less works now because Gnosis Pay has <1000 active users.

Of course, these same "emergency" situations are ripe for exploitation by hackers! I can see future emails or calls from scammers urging people to send their money to a new address. So it's a delicate balance, and there should be safe channels for communication.

3 - There is no way DeFi stablecoin yield compensates users for this level of risk. Some (like me) do it to help move this space forward, but we should have no illusions any sane person will do this without legal guarantees and crypto-specific customer care infrastructure when issues arise.

The current process in crypto is pray the core team will do the right thing... That's not going to scale to a wider audience.

4 - I love Gnosis Pay and will continue to use it, just waiting for word I can recharge my card. A huge thanks to everyone working like crazy behind the scenes. 🙏

[Adam (in SF & enjoying) @adamdbs]

Cool to connect with @ryolu_ on his personal website. There lots of ways new internet communities can form in the age of agents.

[Adam (in SF & enjoying) @adamdbs]

@xdotli @dbabbs @tryprofound @benchflow_ai I respect the library setup

[Adam (in SF & enjoying) @adamdbs]

@simplydt @daoboymatt @daoboymatt chime in sir

[Adam (in SF & enjoying) @adamdbs]

Doing a series on people’s agentic coding set up in SF and what they are building.

@daoboymatt is working on self evolving personal AI. It fine tunes on the fly depending on conversation context. It lives locally and runs multiple models. Agents are persistent and built the entire product.

[Adam (in SF & enjoying) @adamdbs]

Here is the landing page for his lab.

It’s been to see how different founders run their set up. Will do a few more soon.

[Adam (in SF & enjoying) @adamdbs]

He runs 15+ Opus 4.8 or GPT 5.5 agents from his CLI.