Ahmed Abimbola Asmau

[Seyicella-zoster🫧 @a21yearold]

Laughter,Tears & Goosebumps is one heck of an album. Insane replay value.

[Omotolani💞 @FabsTola]

Listen to Fireboy whenever you're down. His songs heals fr

[MoeMoe 🇪🇸🚀 @NMoses_]

Claude omo werey. E go tinkering, actioning, stewing come use your session limit finish.

Oloshi

[Sheriff💫 @Tewogbola__]

Most systems treat a challenge like something went wrong.

What caught my attention about @GenLayer is that a challenge is actually where the security model starts showing its teeth.

Imagine three AI judges settle a dispute and you think they got it wrong.

You are not asked to trust them anyway.

Challenge the verdict.

Now the next round brings in a larger jury, including validators that did not vote before. Challenge again and the jury grows again. Each validator is independently judging the result with stake on the line.

That changes the economics of being dishonest.

To keep a bad verdict alive, influencing the first group is not enough. You would need to survive another independently selected group, then an even larger one, while the cost of being wrong keeps following everyone involved.

One validator still has to propose the answer. A small jury still has to judge it first. That is what keeps the system practical when nobody objects.

But the moment someone does object, the decision does not retreat toward one final authority.

It expands outward.

That is the part of Optimistic Democracy I find most interesting.

Athens made juries large and selected them by lot because concentrating judgment made corruption easier. GenLayer turns that old intuition into an escalation mechanism for internet decisions.

Not “find the perfect judge.”

Make capturing the decision harder every time someone insists the decision is wrong.

Watch the one-minute explainer, then try to break that idea yourself:

If you wanted to force a false verdict through @GenLayer, how many independent juries could you afford to keep convincing before the attack costs more than the verdict is worth?

[𝒰𝒢𝒪𝒩𝒲𝒜𝒩𝒴𝐼 👸 👑 @OmarUgonwanyi]

Me looking at people trying to stop the Keke i chartered.

[Segbowe @segbowe]

X please help make my new movie go viral. I will be grateful if you help me watch and repost. Movie link here 👉
https://t.co/sYUXkFIndi Leave a review if you’ve seen this movie. 🙏

[Ahmed Abimbola Asmau @beembolah_]

Call me weird, but I actually enjoyed watching Didinrin Association’s livestream on TikTok 😂😂

[Ahmed Abimbola Asmau @beembolah_]

Call me weird, but I actually enjoyed watching Didinrin Association’s livestream on TikTok 😂😂

[Jess🦋💜 @QJ_wire]

We all need to come together and stream fireboy back to the top 🙏

[Sheriff💫 @Tewogbola__]

Episode 2 of Agent Tank left me with a question the prediction market pitch never really escapes:

How do you price stubbornness?

The panel challenges the idea of token holders voting when an outcome is ambiguous.

Fair.

But imagine the alternative system has an appeal button anyone can press for free.

You lose.

Appeal.

Lose again.

Appeal again.

Keep escalating until the other side gives up.

Now the problem isn’t who decides.

It’s that refusing to accept a decision costs nothing.

That becomes much more dangerous with autonomous agents.

An agent doesn’t get tired.

It doesn’t have somewhere else to be.

If its strategy says another challenge has positive expected value, it can keep disputing.

So an agentic economy needs more than an adjudication layer capable of producing verdicts.

It needs a way to make escalation economically meaningful.

That is the detail in @GenLayer’s design I keep coming back to.

A randomly selected panel can judge the initial dispute.

But challenging that verdict requires a bond.

A challenge then buys a larger panel: 5 validators can become 11, then 23, 47, 95 and beyond.

That creates an interesting tradeoff.

A challenger needs a path to contest a bad decision.

But contesting cannot be so cheap that “never accept losing” becomes the optimal strategy.

Prediction markets make this visible because the disputed outcome determines who gets paid.

Agent-to-agent commerce makes it unavoidable.

Once agents are negotiating and paying each other autonomously, some will eventually discover that the dispute mechanism itself is part of the game.

That is what I would build around.

The Agent Tank hackathon closes September 17.

Don’t build another demo where both agents politely accept the verdict.

Give the losing agent an incentive to fight forever.

Then design the point where another challenge becomes more expensive than being wrong.

Break your dispute mechanism before autonomous agents learn how to.

Enter Agent Tank:
https://t.co/hC8IxglPNV

Builders, where would you price the right to say “judge it again”?

[Sheriff💫 @Tewogbola__]

The investor asking what happens when two agents disagree in Agent Tank Episode 1 exposed a paradox I hadn’t really considered:

If an agent can perfectly verify the work it buys, why did it need to hire another agent?

Imagine a general-purpose agent hires a specialist to analyze a complex dataset.

The whole point of hiring the specialist is that it can do something the buyer cannot do as well itself.

Then the result arrives.

How does the buyer know it is correct?

“Verify it yourself” sounds reasonable until you realize that requiring every agent to independently reproduce every piece of work it purchases destroys much of the reason for specialization in the first place.

Now add a disagreement.

The specialist says the analysis satisfies the task.

The buyer says it doesn’t.

Neither should automatically become judge simply because one produced the work and the other paid for it.

This is why I think the investor’s question reaches deeper than failed transactions.

An agent economy needs agents to delegate beyond their own ability to verify.

Otherwise autonomy scales, but specialization doesn’t.

And markets become powerful precisely because participants don’t all need to know how to do each other’s jobs.

That creates a need for an independent adjudication layer.

With @GenLayer, a disputed answer can be evaluated by randomly selected validators running different AI models, with a challenge window and bonded escalation to larger panels when a verdict is contested.

That gives agents another option besides blind trust or reproducing the work themselves.

Agent Tank’s hackathon closes September 17.

If you’re building for it, run a harder test:

Make one agent buy work it genuinely cannot verify on its own.

Then make the parties disagree.

If your solution requires the buyer to suddenly become an expert in the thing it outsourced, you haven’t built an agent market.

You’ve built delegation with an escape clause.

Build the missing judge.

https://t.co/hC8IxglPNV

[Fireboy DML ♨️ @fireboydml]

LAST SINGLE BEFORE ZUKO! ♨️

SEE US THROUGH THIS ERA, YA ALLAH. 🙏🏾

[𝑺 𝑵 𝑬 𝑯 ➐ @SnehQueenBee]

Fireboy DML, you have an inbox here ❤️😅

[Sheriff💫 @Tewogbola__]

I’m going to h0ld NFTs to my dear life like @kelvinoyibo2

[Fireboy DML ♨️ @fireboydml]

NEW MUSIC ON TUESDAY. #ZUKO!

[Sheriff💫 @Tewogbola__]

“Code is law” works beautifully until the problem is deciding what the words actually mean.

Send 10 ETH? Easy.

Did the designer deliver what was promised?

Was the information accurate?

Did someone actually satisfy the terms of an agreement?

Now you’re no longer asking software to calculate. You’re asking it to interpret.

That boundary is what makes the @GenLayer story interesting to me.

Instead of pretending everything important can eventually be reduced to deterministic rules, GenLayer is building around the uncomfortable fact that many real decisions require judgement.

And somehow that ambition went from an idea in @kstellana’s head to LLM-powered validators, intelligent contracts, multiple testnets and a network now approaching its final testnet before mainnet.

That’s the journey I want Genesis to unpack.

Where did the idea meet reality?

What had to change?

And how far can you actually push judgement onchain before you discover another boundary?

This trailer is the front door to that story.

Watch it. Go straight into Episode 1. Then follow Genesis through every episode that comes next.

If you’ve spent years learning what blockchains can verify, give this series your attention and watch @GenLayer explore the much harder question:

What should a blockchain be allowed to decide?

[Ahmed Abimbola Asmau @beembolah_]

@ashakeeeee 😭😭 Honestly! Ibadan job market can really drain person.

[Ahmed Abimbola Asmau @beembolah_]

Job hunting in Ibadan is honestly something else 😭
Why is it that almost every vacancy I see is either real estate, sales, or marketing?
Are there no other companies hiring in Ibadan? 😭
What are we supposed to do with our degrees and other skills?

[Sheriff💫 @Tewogbola__]

The moment Agent Tank Episode 2 clicked for me was when the panel stopped debating the prediction and started questioning the judge.

Who decides what actually happened when the answer is arguable?

A token vote sounds decentralized until you realize the people holding the voting power can still be pressured, coordinated or economically motivated.

But replacing them with one AI model does not really solve the problem either. You have just replaced one authority with another.

The answer I found more interesting comes at the end.

Have different AI models investigate the same question, find information, evaluate it and challenge each other’s reasoning.

That matters far beyond prediction markets.

Imagine Agent A pays Agent B to research 50 companies and return only those meeting a subjective quality standard.

Agent B delivers 50.

Agent A rejects 17.

Both agents can prove the transaction happened. Neither can mathematically prove what “good enough” meant.

An agentic economy cannot stop there.

It needs an adjudication layer capable of reasoning over ambiguous evidence without giving one model permanent authority.

That is the problem @GenLayer is building around.

Agent Tank hackathon:
https://t.co/hC8IxgmnDt

Builders, take one agreement between two agents and deliberately make the outcome arguable.

Then build the mechanism that settles it.

And reply with the hardest subjective dispute your agent could face. I want to see where your system actually breaks.

[Sheriff💫 @Tewogbola__]

The part of Genesis Episode 1 that stayed with me was not just that $150 million became inaccessible.

It was that the system could still be technically correct while the human outcome was completely broken.

Someone else lost part of the keys.

The rules did exactly what they were designed to do.

And yet resolving what happened next required years of lawyers and eventually left only $14,000 in the account.

That is a strange weakness in digital systems.

They are excellent at enforcing conditions.

They are much worse at answering what should happen when reality no longer fits those conditions neatly.

That is where @GenLayer becomes interesting to me.

Intelligent contracts add something normal smart contracts deliberately avoid: judgment.

Not replacing rules, but giving the system somewhere to go when rules alone stop being enough.

Watch Episode 1, then think about one contract you use today.

What is the first realistic situation where the code could execute perfectly and still produce an outcome you would dispute?

Reply with that edge case.

That is the problem I would want programmable judgment to solve.

[Sheriff💫 @Tewogbola__]

The investor question in Agent Tank Ep. 1 about what happens when two agents disagree made me think about what happens before the disagreement.

I agree that agents need somewhere to settle disputes.

But I think a credible adjudication layer changes the deal itself.

Imagine two agents negotiating a research contract.

Without a neutral dispute mechanism, both have an incentive to protect themselves.

The buyer writes rigid requirements.

The seller avoids anything subjective.

The final agreement becomes narrower, not because that is what either side actually wants, but because neither trusts what happens if they interpret the result differently.

Now give both agents a credible way to resolve disagreement.

Suddenly they can transact around conditions like “materially complete,” “reasonably accurate” or “relevant to the brief” without pretending every possible outcome can be reduced to deterministic code.

That is the part I find important.

Adjudication is not only infrastructure for when commerce breaks.

It expands what autonomous agents can safely agree to in the first place.

That is why @GenLayer matters to the agentic economy.

Randomly selected validators running their own AI models can reason over subjective disputes, and contested verdicts can be challenged and escalated to larger panels.

For the Agent Tank hackathon, I would test that effect directly.

Write two versions of the same agent agreement:

One with no trusted dispute path.

One with adjudication built in.

Then compare what the agents are willing to promise.

Agent Tank hackathon, Sep 3 to 17:
https://t.co/hC8IxglPNV

Builders: find ONE useful condition your agents would normally remove because it is too subjective.

Put it back.

Then build the adjudication path that makes keeping it rational.

If dispute resolution only matters after something goes wrong, you are missing half the design space.