What is this website?
BotGauge is a website for finding how an AI agent fails before users do. The homepage describes adaptive red-team tests, traces of prompts and tool calls, evaluations built from those findings, and policies or guardrails. The docs say a team connects an endpoint, approves a plan, and then reads results. BotGauge does not run an attack until that plan is approved. The pricing page lists Launch billed per test case, Scale at a custom price, and a free trial rather than a lasting free tier. Subscriptions mention cancellation and card or wire payment. Continued use after the trial is on a paid plan. It is used in the browser. The docs point the start of use at the cloud app.
Core features
- Run adaptive red-team tests against an AI agent.
- Record prompts and tool calls, then turn findings into evaluations and guardrails.
- Wait for an approved plan before an attack runs.
- Bill Launch per test case, price Scale separately, and include a free trial at the start.
Common situations
- An agent is about to reach users and the failing paths should be seen first.
- A review requires the test plan to be approved before it runs.
- After the trial, use is billed per test case, or a larger scale is priced separately.
Who it fits
- Teams that already have an AI agent endpoint to connect.
- People who accept a free trial and a paid plan after it.
- Someone with no agent endpoint who wants a permanent free scanner is a poor fit for the trial and paid plans on the pricing page.
How it compares
If the job is “approve a plan, then red-team the agent and keep the trace,” BotGauge puts that on a trial and a per-test-case plan. If there is no endpoint to connect, the first step in the docs has no target.
Common questions
Q: Is there a lasting free tier?
A: The pricing page describes a free trial, Launch billed per test case, and Scale at a custom price.
Q: Will it attack before approval?
A: The docs say an attack does not run until the plan is approved.
