An AI agent that walks users through setup: one onboarding with Obi by Cor

One customer's onboarding, start to finish

For example, a new customer of a B2B scheduling tool needs to connect their calendar, import their staff list from their HR system and set up their first booking page. Here is how Obi takes them through it.

1. Plan. Obi builds a custom plan for this customer: calendar first, then staff, then the booking page.
2. Reach out. Obi reaches out to the customer directly to start onboarding. Nobody on your team has to chase.
3. Confirm the goal. On the first live call, Obi asks a few questions. Which calendar do they use? How many staff do they have? Who takes bookings? Then it adjusts the plan.
4. Guide. The customer shares their screen. Obi points them to the calendar settings and talks them through connecting it.
5. Follow them out of the app. The staff list lives in their HR system. Obi follows them to that tab and helps them export the right file.
6. Set it up for them. Mapping the file's columns is slow. Obi offers to do it. The customer says yes, and Obi does it inside the customer's own permissions.
7. Clear a blocker. The import rejects some rows because of a date format. Obi sees the error on screen, explains it in plain words and helps fix the file.
8. Hand over when needed. Connecting the company calendar needs an admin, and the customer is not one. Obi explains why, and hands over to your team with what was done and what is left.
9. Keep the plan going. The booking page is still on the plan, so Obi picks it up with the customer in a later session.
10. Report. After each session, your team gets a short report: what is done, and the one step that needs an admin. Nothing else for them to do.

One plan, run from first contact to finished setup, with your team brought in only where a person is needed.

Troubleshooting onboarding blockers

Most blockers are small and specific. A field is in the wrong format. An integration needs a permission the user does not have. A setting sits on a screen the user cannot find. Some blockers sit outside your product, in a CRM, a spreadsheet or an accounting tool. An agent can only fix them if it sees what the user sees.

Obi sees the customer's screen, in your product and in their other tabs and apps. It explains the cause in plain words and fixes it with the customer. If it cannot answer from your help docs, it says so instead of guessing and flags it for your team. When a blocker needs a person, Obi hands over with what the customer was doing, what they tried and where they got stuck, so nobody starts again from the beginning.

How this helps activation

Activation is the moment a new user first gets real value from your product. Most users who never activate stopped during setup. An agent helps activation when more users finish setup, not when more users chat with it.

Obi keeps each customer's plan going until setup is done, and brings in a person only where one is needed. To check it is working, test it on a group of new users and compare their setup completion and activation with a similar group from before. Our guide on how to improve activation rate with AI covers how to define and measure activation.

Automating onboarding with an AI agent: what stays human

An agent can run the onboarding itself: the plan, the outreach, the guided sessions and the setup steps. Strategy, unusual accounts and anything that needs an admin stay with people. Obi hands those over with context. Teams using Obi see up to 2x faster onboarding. For the wider topic, read our guide to customer onboarding automation.

What to check before you pick one

Ask each vendor these six questions, and ask them to show you on your own product.

• Does it build a plan for each customer? Obi: yes, a custom plan per customer, based on what they need to get working.
• Does it reach out first, or wait? Obi: it reaches out to the customer directly to start onboarding.
• Does it see the real screen, inside and outside your app? Obi: the customer shares their screen, and Obi sees their real account and their other tabs and apps.
• Can it set things up, and can you choose which steps? Obi: yes, inside the customer's own permissions, and it asks before major actions. You decide which steps it may do.
• Does it learn from your docs, with nothing to build? Obi: it gets ready from your help docs, onboarding calls or team notes. There are no tours or flows to build. Your team runs Cor from Cor's platform or from inside Claude, ChatGPT and other AI assistants, so nobody learns a new tool.
• Does it hand over with context and report on every session? Obi: it hands over with what was done and what is left, and your team gets a report after every session.

For the full set of tests, see best AI onboarding agent for B2B SaaS. For a shorter example, see how an AI onboarding assistant runs a first setup. New to the category? Read what an AI onboarding agent is, or how an AI onboarding agent differs from a chatbot.

See how Obi works