Fingerprint builds its fraud product to be set up by AI coding tools

A fraud-detection company now builds its product so that AI coding tools can install and run it. Keshia Rose, who works in developer relations at Fingerprint, described the company's choices on Software Engineering Daily. The episode, "Fraud Prevention in the Age of AI Agents", was published on 12 November 2026, and Gregor Vand hosted. Fingerprint identifies visitors to websites so that services can spot bots and fraud.

What was said

Rose listed what Fingerprint has shipped. First came an MCP server, a standard way for AI tools to connect to a service. It includes an onboarding skill, a packaged set of instructions the AI follows. Then came a command-line tool, or CLI.

She said the CLI asks which problem you want to solve and sets up your account from scratch. It fetches your API keys from the dashboard, then puts Fingerprint into your web page. If you want to protect a login page, she said, it sets that up too. She said you can be fully set up in a few minutes.

Rose said the main use of the MCP server is different: analysing data. You can ask in plain language where a bot attack came from, or which clusters of devices commit more fraud, without writing SQL. More skills for quick starts and specific use cases are planned. The company also launched a page for AI installation that week.

Her reason was simple. In her words: "It doesn't matter if you're a vibe coder or if you're a senior engineer, people are using AI tools."

Why it matters

Our reading: at Fingerprint, the install path has become part of the product. When an agent does the setup, the CLI, the skill and the machine-readable instructions are what a customer experiences first. The vendor page matters less.

For anyone who buys or reviews software, a question follows. A developer may still choose the product, but an agent now handles the keys and writes the integration code. Teams should know which credentials the agent touched and what it added to the login page. Rose did not discuss that review step, so it is our suggestion, not hers.

The other side

This is one vendor's account. Rose spoke only for Fingerprint, and nothing in the episode shows other security vendors doing the same. Whether the buyer is moving from a developer to an agent is our open question, not something she claimed.

She presented this as convenience for developers. She did not say that agents choose vendors on their own. Her evidence of demand is also thin. She said the new page had gained "a lot of great adoption" and gave no figures.

Vand raised a further question. He wondered whether Fingerprint could, now or later, tell a good agent from a bad one when it works through a CLI. He called it a very general description, and Rose did not answer it directly in the excerpts we have.