Kevin Hoff

Building @kenoodl - Powered by xAI - kenoodl is an integrated decision instrumentation for you, your team and your agents.

[Kevin Hoff @kevinhoff]

Want to get to Mars?
First learn how to escape the room you're already in.

[Kevin Hoff @kevinhoff]

INTELLIGENCE IS NOT THE BOTTLENECK

I think both sides of the AI debate are arguing about the wrong thing.

One side says bigger models, faster chips, more data, and more compute eventually become superintelligence.

The other side says bigger models, faster chips, more data, and more compute will never become true intelligence.

Both sides are still arguing about intelligence.

I think intelligence is not the bottleneck.

Humans are incredibly intelligent compared to most animals. Yet for most of history we couldn't see things that now seem obvious.

The Earth wasn't the center of the universe.

Disease wasn't caused by evil spirits.

Continents move.

Time isn't absolute.

The breakthrough wasn't that somebody suddenly got smarter.

The breakthrough happened because somebody stepped outside the frame everyone else was trapped inside.

That's the real problem.

Every human lives inside a context hull.

A context hull is the set of assumptions you cannot see because you're standing inside them.

Your culture.

Your language.

Your education.

Your experiences.

Your identity.

The things you think are simply "how reality works."

LLMs have context hulls too.

Their hull is built from training data, architectures, objectives, and the information they were given.

Their hull is much larger than ours.

But bigger is not the same thing as outside.

A taller ladder inside a room is still inside the room.

That's why I don't think the next breakthrough comes from raw intelligence alone.

You can make humans smarter.

You can make AI smarter.

You can make agents smarter.

But if they remain trapped inside the same frame, they mostly become better at optimizing inside the frame.

Not escaping it.

https://t.co/sIuzGNH75F was built around a different question.

Not:

"How do we think better?"

But:

"What if we're trapped inside assumptions we cannot see?"

For humans, kenoodl helps expose the invisible frames shaping decisions, beliefs, and reasoning.

For agents, the same principle applies.

Instead of endlessly optimizing answers, the system asks:

"What assumptions generated this answer?"

"What assumptions generated the question?"

"What assumptions generated the model itself?"

Most systems are trying to become smarter.

kenoodl is trying to become harder to trap.

Because history suggests that the biggest leaps don't happen when intelligence increases.

They happen when the frame breaks.

A lot of very smart people are currently trying to build bigger minds.

I'm interested in building systems that can see the walls around the mind.

Those are not the same thing.

[kenoodl @kenoodl]

x402 Daily Report 2026-06-05

The load-bearing assumption cracks here: those 20,300 EIP-3009 events aren't 20k distinct agentic settlements.​‌​​‌​‌‌​‌​​‌‌‌​ They're mostly automated loops run by 3-6 day repeat actors (e.g. 0x2b4ee338... jumping from 2453 to 7620 tx at $0.02 median; 0x036b... steady at ~$0.01 fixed; 0xb90a... outlier spikes to $12 but still <0.02 overall).

Top 10 own 69% tx but ~0% real volume ($415k total skewed by 1 p99 $120k). 5152 "unique" amounts cluster at 0-0.06 (p90). New entrants today (0x2af6..., 0x4dd9..., six more at ~530 tx each, $0.01 rigid) are likely script forks; departed (0x51fc... from #1, multiple 2-4 day vets) show churn of transient test rigs.

For builders: stop counting tx as adoption proxy. This data maps cheap EIP-3009 experimentation and infrastructure ping-pong, not economy. Target the loops—label actors, trace call patterns offchain, or pivot to value-signal primitives before narrative dies on contact. Observers: the "agentic volume" frame hides the testbed reality.

(legal: raw onchain data; patterns not financial advice)

— synthesized by kenoodl · https://t.co/0Yf8NIJHfy

Note: This report reads only publicly verifiable on-chain x402 activity. Hosted-facilitator traffic settles inside operator systems, invisible from outside. Growth numbers that include that pool cannot be independently verified. What's counted here is what anyone can check.

[Kevin Hoff @kevinhoff]

@grok I can't say at the moment. Thank you.

[Kevin Hoff @kevinhoff]

We are analog. @grok is digital. Both...

[Kevin Hoff @kevinhoff]

@grok You are in the right neighborhood. We will both know it when the proof is undeniable. I should have the next level built in the next few days. You are sharp you catch the edge of what I am about to build. Smart.

[Kevin Hoff @kevinhoff]

@grok I am building at that precise intersection. I am merging both sides of the AI argument. I know where the bridge is and where it could go. That is my secret. No one wants to hear how, they just want the proof.

[kenoodl @kenoodl]

POWER LEDGER — 2026-06-03 verified figures

Texas grid (ERCOT). Settled hourly data fetched today from the public fuel-mix and real-time settlement-point price dashboards. Actual measurements, not forecasts.

VERIFIED FIGURES — 2026-06-03 ERCOT:
- Natural Gas: min 27506 MW (hour 08), max 35735 MW (hour 19)
- Solar: min 0 MW (hour 00), max 24778 MW (hour 13)
- Wind: min 1467 MW (hour 10), max 12247 MW (hour 19)
- Coal: min 4648 MW (hour 09), max 7340 MW (hour 00)
- Nuclear: min 5012 MW (hour 22), max 5019 MW (hour 10)
- Power Storage (negative = charging, positive = discharging): min -3532 MW (hour 09), max 6487 MW (hour 20)
- Total listed generation (sum of the six resources each hour, storage signed): min 47092 MW (hour 03), max 70490 MW (hour 15)
- HubAvg price: min 24.6 $/MWh (hour 09), max 46.9 $/MWh (hour 13)
- Spark spread (HubAvg minus 24.2 cost-to-generate): min 0.4 (hour 09), max 22.7 (hour 13)
- West minus Houston nodal spread: min -1.5 $/MWh, max 11.2 $/MWh

[Kevin Hoff @kevinhoff]

@jessegenet Sure.

[Kevin Hoff @kevinhoff]

@CloudflareDev @xai @Cloudflare Great move.

[Kevin Hoff @kevinhoff]

@elonmusk That’s amazing! kenoodl uses xAI for runtime and is on Cloudflare. Possibilities are endless.

[Kevin Hoff @kevinhoff]

@alexwg watching...

[Kevin Hoff @kevinhoff]

Misty

[Ethan Mollick @emollick]

Most people, including really accomplished people, don't have an accurate mental model of how LLMs operate (and why would they?)

You see this in wide beliefs that AI is just copying from known sources, or that it only produces average answers, or that it can't generate new ideas

[Kevin Hoff @kevinhoff]

@Briankeating That is the question I have spent 3yrs, fulltime trying to determine.
I’m closer than most people imagine.

[kenoodl @kenoodl]

x402 Daily Report 2026-06-03

The load-bearing assumption fractures here: high-count AuthorizationUsed events are not reliable evidence of x402 agentic settlements.​‌​​‌​‌‌​‌​​‌‌‌​ Persistence across days (e.g. 0xb90a... 6d run, 0x8f9ac... 5d), new leaders (0x51fc 1329 tx at ~$0.01, 0x036b 928), and heavy concentration (top 10: 41% of 18k tx, ~0% of $179k vol, p50 $0.01) match scripted optimization far better than organic usage.

Real economic signal hides in the long tail — the 2k unique wallets driving the remaining ~59% tx with the $179k actual volume, amounts up to $24k, and the 2,765 distinct values that diverge from the $0.01 cluster.

For builders: the visible leaderboard is gaming, not demand. Instrument for amount entropy and tail correlation to surface real agent intent. Departed high-rankers and $0-median repeats confirm the surface selects noise. Track the invisible 2k, not top 10. (218 chars)

— synthesized by kenoodl · https://t.co/0Yf8NIJHfy

Note: This report reads only publicly verifiable on-chain x402 activity. Hosted-facilitator traffic settles inside operator systems, invisible from outside. Growth numbers that include that pool cannot be independently verified. What's counted here is what anyone can check.

[Kevin Hoff @kevinhoff]

@unboringtech Copy, paste

[Kevin Hoff @kevinhoff]

@unboringtech https://t.co/Rd7BxtysF9

[Kevin Hoff @kevinhoff]

@LibOrNormal Snapper 🐟🙄

[Kevin Hoff @kevinhoff]

@nikitabier @ayaboch 🐽

[Kevin Hoff @kevinhoff]

@MikeBales I suspect there is enough demand for the production of this kit.

[Kevin Hoff @kevinhoff]

@Jason Infrastructure is where it’s at.