Co-founder @Britebot70: AI agents that do real work. Founder @ThinkForesight: See around corners in tech & money. President @TennesseeBTC. ✍️ Forbes + elsewhere
Tennessee, USA
Joined September 2021
- Tweets2.6K
- Following856
- Followers1.2K
- Likes71.7K
Pinned Tweet
Most "AI agents" are demos. We build the boring kind that actually ships.
At Britebot, agents answer support requests, follow up on leads, book appointments and move data between systems, on our infrastructure or yours, with local models when data can't leave.
brite.bot
Dave Birnbaum retweeted
It's still such an adorable moment when the agent casually clicks the "I am not a robot" checkbox
Another way AI is surpassing human capability. They’ve been coming fast and furious all week.
> AI now catches far more errors in clinical trial papers than human peer reviewers: all 15 models tested scored 53% to 96% on planted errors, while untrained human reviewers averaged just 27%.
> The result held on 58 new trials published in 2026, so the models weren’t simply recalling old papers they had seen in training; Claude Opus 5.5 scored highest at 99%.
> Humans miss the errors that most change a trial’s meaning, catching text-table contradictions only 7% of the time, and they take 5 to 8.5 hours per review against AI’s 1.4 to 16.3 minutes.
Dave Birnbaum retweeted
If your agents are running but nobody knows what they're spending, that's the first thing an audit surfaces. brite.bot/business
Everyone is freaking out that the 512GB Mac Studio costs over $22,000. But a top-spec Quadra 950 was $9,199 in 1992, about $21,500 in today's dollars.
The Quadra shipped with 8MB of RAM.
Satellites may soon launch in pieces and build themselves in orbit. I see big applications here for space based solar.
Rendezvous Robotics has raised $20 million (led by MaC Venture Capital and Toyota Ventures) to develop self assembling spacecraft.
The system uses roughly 1 meter hexagonal tiles with sensors and onboard processing. The tiles lock together, letting a satellite launch flat, assemble in space, and later reconfigure for repairs or upgrades.
The technology builds on nearly a decade research at MIT. Earlier versions have flown on zero-G aircraft, a suborbital mission, and two ISS flights. The company plans an in space demo of seven tiles no earlier than April 2028, after sending 32 tiles to the ISS for further testing.
The approach aims to escape rocket fairing size limits and enable larger, more adaptable satellites without astronauts or robotic arms.
Progress (and spacecraft that put themselves together) marches on. 🛰️🔧🚀
Source teapawt.org/payloadspace/status/21…
High-intelligence AI is safer than low-intelligence AI.
Cyanobacteria didn't outthink anything, they just copied themselves until the planet changed.
Low intelligence means geometric proliferation is your only leverage. A smart agent has cheaper options than eating the world.
If you are worried about gray goo, you should love superintelligence.
The Soviet flag psyop theory is remarkable.
The self-described "noticers" never noticed a star on a flag they have seen a thousand times?
Then instead of being mildly embarrassed and staying mum, they proclaim a planet-wide psyop. Or maybe a "time shift"? Who knows.
The embarrassment is totalizing.
A new Mandela Effect is rising.
Since when does the Soviet flag carry a star above the hammer and sickle?
Many of us remember it clearly without one.
If you recall the flag without the star… you may be from the old Earth.
Shifted into this denser, clown-world timeline.
The memory itself is the proof.
The body knows.
The soul knows.
We are not crazy.
We are remembering.
I miss the old Earth too.
Yet we are here now to hold the Light and anchor a higher one.
Who else feels this?
This is presented as negative but 6.4 million agent payments of about 1.9 cents each is actually incredibly bullish for agentic micropayments.
Almost everything you hate about the internet exists because the web never learned to settle a 2-cent payment. So it sold your attention.
And if it actually works you would expect the ratio of transaction volume to transaction size to be small and getting smaller!
➥ 6.4 million AI agent payments, but only $120K moved
that's the reality of x402 payments today, and i think it's time we look beyond tx counts and focus on the actual economic value AI agents are creating.
according to BeInCrypto's latest research, here's what happened over a 35-day period:
– 6.4M x402 transactions processed just $119,947.
– 90.8% of payments were worth less than $0.01.
– avg tx value was only \~$0.019.
– institutional readiness scored just 12/100.
– 14 out of 19 AI tools still operated under human-defined spending limits.
clearly, AI agent payments are gaining adoption, but meaningful economic activity is still limited.
most transactions involve API calls, data access, and other tiny automated payments.
i don't think micropayments are a bad thing. they're actually a perfect use case for x402.
but there's a big difference between an agent spending $0.01 on an API call and a business trusting one to manage $10K.
so what needs to improve before AI agents can handle real capital?
[1] Agent Identity & Trust
businesses need to know which agent they're dealing with, who controls it, and whether it can be trusted.
[2] Spending Controls & Risk Management
agents need programmable spending limits, fraud detection, permission systems, and compliance checks before executing transactions.
[3] Real Economic Demand
we need more agents paying for actual services, B2B operations, trading, compute, and automated financial workflows.
and if this happens, i think several projects are worth watching:
→ solana:PAYmo6moDF3Ro3X6bU2jwe2UdBnBhv8YjLgL1j4DxGu | @PayAINetwork: x402 payment facilitator handling verification and settlement for AI agents.
→ $KITE | @GoKiteAI: building identity, payment infrastructure, and programmable permissions for autonomous agents.
→ skale:native | @SkaleNetwork: supporting gas-free x402 micropayments and pay-per-use API marketplaces.
→ $VIRTUAL | @virtuals_io: developing an agent economy where AI agents can provide services and transact with each other.
→ $BNKR | @bankrbot: building financial infra, wallets, and tools for agents to manage funds and execute on-chain actions.
→ t54 | @t54ai: building identity verification, risk assessment, and credit infrastructure for agent payments.
i'm still bullish on the agent economy, but i'll be watching real payment volume, revenue, and token value capture much more closely than raw tx counts.
Incredible story about how cheap intelligence shrinks the firm to the minimum size needed to exercise good judgment.
That’s in contrast to a corporation that pools many brains into one bigger intelligence at the price of sky-high overhead.
One founder plus Claude just took a drug from concept to mouse data in 18 months, then sold to Anthropic!
Sam Altman's one-person company prediction is coming true. Anthropic just bought one. The founder of Mithril Medicine used Claude to take a muscular dystrophy antibody treatment from concept to proof-of-principle in mice in 18 months, with just himself, Claude, and part-time consultants. The technical report is great. I will quote:
'The week opened with the handover of two cell-based assays, cell adhesion and flow cytometry, within our team, while eleven questions remained open with the assay CRO. In two sentences, one of us asked Claude to review the latest emails between the outgoing assay lead and the assay CRO and to recommend next steps for the handover, and gave it the link to the handover folder. Claude returned a bottom line after 96s and a full plan with five drafted emails after 18 min. For the full plan it read six email threads, the handover folder, two slide decks and two reports, all in the mailbox and file store to which we had given it access. Four working days later, nine of the eleven questions had an answer in writing and the other two had been left to pilot experiments. The adhesion screen was thereby set out in writing. The CRO drew the plate map to the format that our team had requested, Claude compiled the run order and the material calculation from the production data, and one of us reviewed and approved them.
On day 4, the production CRO delivered its data package, a report of 341 pages, which one of us forwarded to Claude in two sentences, with the request to review it, share observations and recommend next steps. Claude returned a bottom line after 2 min 40s and, once it had read every page, a visual synthesis of all 113 constructs after 7 min 53s. The synthesis arranged the constructs on the grid of 14 by 8 arms. Claude marked 82 constructs for screening as is, seven for additional polishing, and the 24 constructs of three laminin-binding arms to be held back from the cell-based assays on production criteria. Each of the 113 constructs thus had a proposed next step in less than 8 min.
Claude had recommended binding kinetics for all 113 con-structs, and on day 7 one of us challenged this plan. Claude defended it within 27s, arguing that affinity would show whether the three arms merit further work. One of us overruled this recommendation and set the three arms aside on production criteria. Their material remains in storage. Four minutes later, Claude had revised its memo, the working page, the run order and three unsent drafts from 113 to 89 constructs. It had also applied the ruling to three dependent steps in the unsent drafts, each flagged for our veto. The exchange took 7 min 36s from challenge to new plan, and every affected email was still an unsent draft. After the ruling, eight emails drafted by Claude went out on the same day to three CROs and a consultant.
Over the whole week, Claude drafted 17 emails, each in answer to a request from one of us, and we sent 12 of them, each after editing. With its drafts, Claude listed 41 points, such as tube fills, shipping conditions, recipients and assay settings, for which it had found no written statement of ours, so that we could confirm, change or strike each before sending. We kept some, struck others and changed one. Among the passages that we kept were proposals for controls. One draft proposed a comparison of polished and unpolished material on one sensor chip, to size the error that impurities cause in the rest of the panel. Another requested a buffer control diluted like the most dilute sample, so that any effect of the buffer appears at its largest. These proposals concerned the measurement of constructs that had already been produced, and each reached a CRO only after one of us had reviewed and approved it.
Communism is responsible for the death of at least 100 million people and human suffering on an unimaginable scale. To argue that the CCP isn’t truly Communist is one thing… but to try to rehabilitate Communism itself is sick.
grokipedia.com/page/The_Blac…
Dave Birnbaum retweeted
@Microsoft's Dataverse MCP Server now supports custom APIs, so AI agents can invoke existing business logic as a single atomic transaction on the server, regardless of where the agent is hosted. If any step fails, Dataverse rolls back everything.
We know single individuals can now run entire software companies. What about construction sites? Imagine buying some land and vibe-building the family compound of your dreams…
Today we’re announcing a $600M Series C, co-led by @CapitalG and @emergencecap, to bring fully autonomous machines to more job sites across the U.S.
We want autonomous equipment to become beautifully unremarkable, as ordinary on a job site as GPS or grade control.
Dave Birnbaum retweeted
Stop trying to "lock in" and start seeing work as play.
The results will follow.
Apollo deniers always use this flawed logic. The simple answer is, landing people on the moon is not as hard as it sounds.
Hermetically sealed containers are everywhere. Do you deny soda cans and scuba tanks? If not, you concede it’s possible to seal off some
dudes in a metal can filled with air from oxygen tanks.
Rockets are everywhere. You can buy a model rocket and watch it fly. You can shoot a gun. (You can also watch rocket launches, which your own eyes would tell you are real, at least until the rocket disappears into the sky.) Do you deny that a chemical reaction could accelerate a projectile? If not, then you concede large rockets are possible; just scale up what you experience when you go shooting.
Newtonian mechanics are visible everywhere. Throw a football with controlled power and aim, and the football will always land in the same place. Do you deny that a quarterback can predict how to throw a football so that it intersects with a particular time and place? If not, then you concede you can use simple math to make a projectile collide with the moon.
The devil is in the details, of course – it’s hard to make the whole thing work with 99% chance of success. But there’s nothing about visiting the moon that pushes the limits of science.
Dave Birnbaum retweeted
A hole on the Moon just turned out to be the front door to a 50 km tunnel.
Japan’s Kaguya orbiter first spotted it. A pit about 50 metres wide and 50 metres deep, sitting in the Marius Hills on the Moon’s near side.
Then JAXA pointed the probe’s radar underground. The echoes came back with something huge: a lava tube roughly 100 metres wide, running about 50 km, and still intact. No collapse.
Scientists think it was carved by volcanic activity around 3.5 billion years ago. There may even be ice or water locked in the rocks inside.
That’s why JAXA thinks it could become a lunar base. Shelter from radiation, micrometeorites and brutal temperature swings, all built in.
Source: JAXA / ISAS, Geophysical Research Letters
Dave Birnbaum retweeted
The funny thing is everyone was saying AI slop code will cause a cyber crisis but it’s actually all the old human slop causing it and the AI is just cleaning it up
