Web3AI Agents · Web3Anonymized · NDA

Cipher

On-chain analytics with an AI agent that monitors wallets, explains flows, and alerts on anomalies in plain language.

Web3 analytics company (name withheld, NDA) · 14 weeks, design to launch

Client
Web3 analytics company (name withheld, NDA)
Sector
Web3
Team
5-person team
Timeframe
14 weeks, design to launch

The problem

A Web3 analytics company had rich on-chain data and a product only power users could read. Analysts watched wallets by hand, and explaining a suspicious flow to a client meant an expert, a whiteboard, and an hour.

Constraints

On-chain data volumes ruled out naive polling, and alerts had to be right; false positives would burn the client's credibility with their own customers. Every explanation the agent gave had to cite the transactions it was based on.

What Trace built

An analytics platform with an AI agent that monitors watched wallets, explains fund flows in plain language with citations to the underlying transactions, and alerts on anomalies. Tuned thresholds, an evaluation set built from real cases, and human review on high-severity alerts.

Team

1 product designer2 engineers1 AI engineer1 engagement lead

Timeframe

14 weeks, design to launch

Outcome

Wallet monitoring that ran on analyst hours now runs continuously, and every alert arrives with a plain-language explanation a non-expert can act on. Detailed metrics are under NDA; this study is shared with the client's permission, anonymized.

Questions

About this project.

What did Trace build for this client?

An on-chain analytics platform with an AI agent that monitors wallets, explains fund flows in plain language with citations to the underlying transactions, and alerts on anomalies.

How did you keep the AI agent reliable?

Every explanation cites the transactions it's based on, alert quality is measured against an evaluation set built from real cases, and high-severity alerts go through human review before they reach a customer.

Why is this case study anonymized?

The client operates under strict confidentiality obligations. The study is shared with their permission with identifying details removed; we're happy to walk through the engagement privately on a call.

Next case study

Ledger

Your project

Building something with similar stakes? Tell us about it. We'll say plainly whether we're the right team.

Book a call