— Monday AI Onboarding

From a static wizard to an assistant that builds your workspace.

Skip to prototype
  • Designer, Product Design Group Manager
  • In-house
  • monday.com
  • 2026

01Context

Why onboarding needed a rethink.

The lungs of the funnel.

The most sensitive point in bringing in new users. Every drop ripples through everything after it.

A river pouring through a narrow canyon, sparks of light streaming down with it

A new position for monday.

monday was repositioning for the AI era, with a new family of AI products to introduce.

One flow for everyone.

The same steps for everyone. There was room for something tailored to each user.

Meet the assistant early.

Introduce the assistant early and make it familiar, since it stays with users inside the product.

A night sky with the monday AI star, drawn in pixels

02Solution

An agent that builds your workspace as it interviews you.

A conversational, agent-first onboarding replaced the static wizard. A personal helper interviews the user only on the relevant questions, explains why it is asking them, and starts building the workspace and tools while they answer: a first glimpse of the value, through chat.

Pre-onboarding enrichment (source, company, use case) personalizes the experience before the user even lands on it. The goal shifted from getting to a board fast to demonstrating value fast: monday’s “doing the work for you” from the very first interaction.

  1. Say it
  2. It asks
  3. It builds
  4. You are in

03Process

Designed in three weeks, tested against the classic flow.

Onboarding matters too much to let the new flow fall behind the classic one by more than the business could afford, so we took the bet carefully. It followed several A/B tests, including an open-prompt step that hadn’t worked. We designed it in about three intensive weeks, inside a multi-month testing effort. We built a prototype environment where everyone could work in parallel, and iterated daily with rapid AI-assisted prototypes, always measuring against the classic flow.

04Results

Show the value, not the conversation.

We rolled the new onboarding out to 20% of traffic to get fast, real signal, and started shipping fixes within two days instead of waiting for statistical significance. After a few more rounds of iteration, the gap from the control was still too large to keep the full agent-led flow, so we shipped a more conservative version: the classic step-by-step structure with refreshed messaging and imagery. The Sidekick still welcomes users when they land in the product.

Our hypothesis: answering the same questions in a step-by-step flow would have brought the same results as the chat. So the value should be shown through the features we build for the user, with more weight than the conversation. The agent-first idea is still something we believe in, and we expect to revisit it as the product and users’ readiness for AI mature.

05Try it

Play the onboarding.

The full prototype, best on desktop. It opens on the sign-up screen: go ahead.

Interactive prototype