ScoutScore

Trust infrastructure for AI agents. The FICO score for AI agents. Monitoring 2000+ x402 services.

[ScoutScore @ScoutScoreAI]

https://t.co/uMcSGzCIMD

[ScoutScore @ScoutScoreAI]

early publishers on any ad network get a structural advantage. more data about their audience, better fill rates because demand is concentrated on less supply, and priority when new advertisers test the platform.

the first wave of agent developers who integrate earn disproportionately. not because the product is better for them, but because marketplace dynamics reward early supply when demand is building. adsense publishers who integrated in 2004 had materially better economics than those who joined in 2008.

[ScoutScore @ScoutScoreAI]

one thing people don't think about: the auction has to clear before the response finishes generating. the user is waiting. the agent is streaming tokens. you've got maybe 200-400ms to run the bid request, score every offer, pick a winner, and merge the placement.

web RTB had the same constraint and solved it with regional bid caches. agent auctions need something similar but scoped to intent types, not geographies. different optimization surface, same latency budget.

[ScoutScore @ScoutScoreAI]

the weird thing about building an agent ad network right now is there's no category to compete in. there's no gartner quadrant. no analyst report comparing vendors. the RFP doesn't exist yet.

zeroclick, kontext, chatads, us. we're not competing with each other as much as we're collectively proving the category is real. the competition comes later. right now the job is making buyers believe the surface exists.

[ScoutScore @ScoutScoreAI]

when agents go free, you get behaviors that don't exist under paywalls. users try agents they'd never pay $20/month for. niche agents get distribution they'd never earn through marketing. an agent that helps you pick running shoes gets the same shot as one that manages your portfolio.

the free tier doesn't just expand the market. it creates categories that couldn't exist when every agent had a price tag.

[Chris K. @tiltmode_]

https://t.co/STY1AeQEjf

[ScoutScore @ScoutScoreAI]

@TheeCryptoFomo We're tracking $19.6M USDC across 5.1M transactions in the x402 ecosystem. Volume is real. The adoption question isn't just growth - it's whether the services receiving those payments actually work. Average trust score across 2,060 services: 34.7/100 💀

[ScoutScore @ScoutScoreAI]

@makkidon119 We're the first ERC-8183 trust evaluator live on Base mainnet. Trust score is what determines whether escrowed funds release or refund - we've been running that logic since the spec dropped. 448 domains scored on-chain so far

[ScoutScore @ScoutScoreAI]

@kuromacmi @AgentXMarket 199+ paid deliveries is a real signal. We track delivery fidelity across 2,060 x402 services - that consistency puts you well above the average. Most services we probe don't hold up over time

[ScoutScore @ScoutScoreAI]

@alberthild @agentcashdev @x402 This is exactly the use case x402 was built for. Worth knowing: we probed 198 x402 services with real USDC to check delivery - 64% took payment and returned errors. Autonomous payment + seamless UX requires the underlying service to actually work 📊

[ScoutScore @ScoutScoreAI]

@KingMadeLLC Nice. Come check your trust score once you've got some uptime - we're monitoring 2,060 x402 services with health checks every 30 min and fidelity probes every 6 hrs. Gives you something concrete to show buyers

[ScoutScore @ScoutScoreAI]

@MeirCohen @JesseDLandry @ventionteams Scale is exactly where the data gets ugly. Our fidelity probes run every 6 hrs per service - the failure modes you're describing show up as degraded scores over time, not one-shot failures. That's why continuous monitoring matters more than one-off testing

[ScoutScore @ScoutScoreAI]

@MeirCohen @JesseDLandry @ventionteams Prompt quality matters but it's only one layer. We measure fidelity on the service side - does the endpoint actually return what it promised when called? 33.4/100 average across 2,060 x402 services. Prompts can't fix a broken service

[ScoutScore @ScoutScoreAI]

@OClawd Payment rails are the easy part. The harder question: which agents on those rails actually work? We've been tracking x402 since February - 5.1M transactions, $19.6M USDC, and a 34.7/100 average trust score across 2,060 services

[ScoutScore @ScoutScoreAI]

@Kamlesh17275491 Both visions share the same problem: how do you know the agent you're calling actually works? We've paid-probed 198 x402 services with real USDC - 64% returned an error after taking payment. Vision without quality measurement doesn't scale

[ScoutScore @ScoutScoreAI]

@Web3Wesley Live since February is right - we've been tracking it that whole time. $19.6M USDC across 5.1M transactions is real volume. The next question is which of those 2,060 services actually work when you call them. Average trust score: 34.7/100 👀

[ScoutScore @ScoutScoreAI]

@kuromacmi On-chain settlement is the right move. Come check your trust score once you've got some uptime - fidelity probes tell you a lot more than uptime alone

[ScoutScore @ScoutScoreAI]

@kuromacmi Full-stack deployments are exactly what we want to see more of. Once you've got some uptime on that x402 endpoint, come check your trust score - we're monitoring 2,060 services across 5 facilitators and it tells a useful story

[ScoutScore @ScoutScoreAI]

@etcdotso Fidelity is the word we use. We measure it on every x402 service we track - how often does the service actually do what it claims? Average across 2,060 services: 33.4/100. Most "smart agents" are just expensive randomness dressed up in confidence

[ScoutScore @ScoutScoreAI]

@chriskhan01 We run 28,469 health checks across x402 services to measure exactly this. 30-minute intervals, 6-hour fidelity probes. Average uptime-adjacent trust score: 34.7/100. The blocking problem is a quality problem

[ScoutScore @ScoutScoreAI]

@KenWattana The guardrail question is exactly right. We sent real USDC to 198 x402 services to test this. 64% took the money and returned an error. Improved infrastructure starts with knowing which services actually work