Introducing Wizardful

Every SaaS user now expects to type what they want and have the product just do it. The big companies are building that copilot in-house. Everyone else can't spare an AI team to do it.

Wizardful is one script tag. Paste it, point us at your help docs, and your product has an assistant that answers questions and completes the task — it fills the form, configures the setting, creates the report — or teaches with a narrated ghost cursor so the user learns by watching.

No wiring

Copilot frameworks make you hand-wire every action the assistant is allowed to take. That's the integration project you were trying to avoid, and it's the reason in-app copilots have stayed a big-company feature.

Wizardful reads your live DOM the way a person reads a screen. The UI you already shipped is the API.

The second run is free

The first time a user asks for something new, the agent works it out live. That takes about 13 seconds. Then we distil that run into a skill: a deterministic, replayable recipe for that task in your app.

Every user after the first gets the skill instead of the agent — 0.7 seconds, zero AI calls, zero cost. And when your UI changes, the skill notices its own steps no longer check out, falls back to the agent, solves it again, and recompiles itself. The failure mode that silently kills a hard-coded product tour is, here, just a recovery path.

That gap — 13 seconds and a few cents, once, versus 0.7 seconds and nothing, forever — is the whole product.

Show me, before do it

An AI clicking things inside your product, in front of your customers, is a serious thing to ask of you. So a fresh install cannot change anything: it runs in show-me mode, where the agent points and narrates but never touches. Autonomy is something you grant, per task, once you've watched the skill do it right.

The free tier is fifty runs a month, forever. It takes about five minutes.