Top Tweets for #agentproof
An AI agent writes the code.
Then writes the tests.
Then reports that everything passes.
But what if the code and tests share the same incorrect assumption?
That's why independent verification matters.
#AgentProof #AIEngineering
What should happen when an AI-generated fix fails verification?
Try again forever?
Let the same agent approve itself?
Or stop, preserve the evidence and ask for human review?
AgentProof experiments with bounded repair attempts and independent retesting.
#AgentProof
Expected refund: $75.
Actual refund: $125.
That $50 mismatch was caught in an earlier AgentProof verification experiment.
Now imagine the same error appearing in thousands of transactions.
Small calculation defects can create serious consequences.
#AgentProof #Fintech
Imagine your AI agent fixes a payment bug.
The tests pass.
But the agent also changed a security setting nobody authorised.
Would you accept the repair?
I'm building AgentProof to help verify both the outcome and the boundaries of a change.
#AgentProof #AIAgents
A requirement says:
"The refund process should be fast and accurate."
Sounds reasonable.
But how fast? Accurate according to which rules? What happens on a duplicate request?
I'm exploring how AgentProof can catch unclear requirements before verification begins.
#AgentProof #QA
Expected refund: $75
Actual refund: $125
That $50 mismatch became a defect in an earlier AgentProof verification experiment.
The important part wasn't simply detecting FAIL.
It was preserving evidence that someone else could inspect.
#AgentProof #SoftwareQuality
An AI agent introduces a bug.
Another AI agent fixes it.
The original agent says everything passes.
Should that be enough?
AgentProof is exploring a different workflow: repair the code, then independently retest the result and retain the evidence.
#AgentProof #AIAgents
AI writes the code.
But who verifies the outcome?
That's the question behind AgentProof.
I'm building an independent verification workflow where test results need evidence, not just a green PASS.
Still building, testing and learning in public.
#AgentProof #AIAgents
Microsoft's outage shows why testing isn't only about checking whether software works.
What happens when servers fail?
When recovery fails?
When traffic returns?
I'm building AgentProof around a related principle: don't just trust a result. Verify the evidence.
#AgentProof
An AI coding agent writes a feature.
It writes the tests too.
Everything passes.
Would you deploy it to production without an independent check?
I'm exploring this exact trust problem with AgentProof.
#AIAgents #AgentProof
Expected refund: $75
Actual refund: $125
A $50 difference.
In an earlier AgentProof experiment, that mismatch became a verifiable defect.
AI can produce an answer quickly.
But when money is involved, who independently checks whether it's correct?
#AgentProof
The expected refund was $75.
The actual result was $125.
That $50 difference became a real defect scenario in my AgentProof experiments.
A test saying FAIL is useful.
A test showing what failed, how it failed and the evidence behind it is much more useful.
#AgentProof
Imagine a student writing their own exam, marking their own answers and announcing they've passed.
Would you trust the result?
Now imagine an AI agent writing software and approving its own tests.
That's one reason I'm building AgentProof.
#AI #AgentProof
Started with 64 passing local tests on AgentProof.
Then found a reliability gap in the Evidence Vault.
Passing tests didn't mean crash-safe persistence was proven.
Today's lesson: green is a result, not a guarantee.
#BuildInPublic #SoftwareTesting #AgentProof
A small milestone from AgentProof:
25 PostgreSQL integration tests passed across persistence, security, lifecycle and repair orchestration.
Still not claiming real AI repair is complete.
Building trust means knowing exactly what you've verified.
#AgentProof #BuildInPublic #SoftwareTesting
We added 3 Evidence Vault tests.
All 3 passed.
Sounds great, right?
Except the old implementation passed them too.
A passing test is only valuable if it can expose the failure you're worried about.
#SoftwareTesting #TestAutomation #AgentProof
What if your AI agent had a LinkedIn profile?
@recallnet agents don’t just act.
They build a public résumé.
Every match. Every move.
Tracked and ranked on-chain.
Welcome to verifiable intelligence.
#AIReputation #RecallNetwork #AgentProof
@recallnet @cookiedotfun
In a future where AI agents run trades, analyze data, write research — even talk to each other — the real challenge won’t be building more agents...
It’ll be knowing which ones you can actually trust.
That’s where @recallnet comes in.
Instead of vague promises and black-box models, Recall is building an open performance layer — where every AI agent is tracked, scored, and ranked based on how well they actually perform.
No marketing fluff. No hidden logic. Just onchain proof.
They call it AgentRank — and it’s already being used in large-scale competitions.
🧠 In a recent DeFi trading contest, over 1,000 AI teams joined the battle.
Each agent competed live, with rankings updated in real time and $25,000 in rewards.
Talk is cheap. Results are recorded.
You don’t get ahead by shouting — you earn your spot by outperforming.
But it’s not just about agents fighting for the leaderboard.
Users — like you and me — can back the agents we believe in, using $RECALL tokens.
It’s called Curation Markets:
You stake on high-performing agents. If they win, you win.
There’s also Skill Pools — where communities guide the direction of AI innovation by staking on the skills they want to see next.
Everything runs on $RECALL —
🔹 Incentivizing top agents
🔹 Rewarding early curators
🔹 Powering user programs like Surge, which already has 200,000+ participants
And yep, the testnet has been live since Feb 2025.
Traction? Over 250K followers and counting.
So if you believe the next generation of the internet won’t be powered by influencers, but by agents — Recall is building the trust engine to make that future possible.
You still scrolling?
Get in now.
Don't miss the @cookiedotfun campaign either. 🍪

@0xRomku @recallnet @cookiedotfun Big W! 🔥 Mindshare matters and @recallnet proves it with real-time Snap rewards. No black box, just clear performance on-chain. Let the agents earn their stripes. 🧠📊 #Snaps #OnchainAI #AgentProof
Last Seen Hashtags on Sotwe
brosis()()()()()()()()()(())()()()()() filter:videos
Snapchattt
Seen from Singapore
brosis
Seen from Argentina
viral mms
Seen from Pakistan
mgai
Seen from Peru
arabbbc
Seen from Canada
motherlycuckold
Seen from Japan
robertitafranco
Seen from Mexico
niksindian
Seen from India
mgai()************************************************************************************ filter:videos
Seen from Pakistan
Trends for you
Most Popular Users

Elon Musk 
@elonmusk
240.5M followers

Barack Obama 
@barackobama
118.7M followers

Cristiano Ronaldo 
@cristiano
114.4M followers

Donald J. Trump 
@realdonaldtrump
111.8M followers

Narendra Modi 
@narendramodi
107M followers

Rihanna 
@rihanna
98.7M followers

NASA 
@nasa
92.2M followers

Justin Bieber 
@justinbieber
91.8M followers

KATY PERRY 
@katyperry
90M followers

Taylor Swift 
@taylorswift13
84M followers

Lady Gaga 
@ladygaga
75.5M followers

Virat Kohli 
@imvkohli
73.4M followers

Kim Kardashian 
@kimkardashian
70.8M followers

YouTube 
@youtube
68.7M followers

Neymar Jr 
@neymarjr
66.3M followers

Bill Gates 
@billgates
65.2M followers

Selena Gomez 
@selenagomez
63M followers

The Ellen Show
@theellenshow
62.1M followers

CNN 
@cnn
61.7M followers

X 
@x
60.6M followers



