OnRamp AI on the Customer Portal

Keeping customers moving through their onboarding can be harder than building the project itself. They lose their place, miss updates, and don't always know what to do next — or why.

OnRamp AI is built into the customer portal to change that. This article covers what it does, how your customers experience it, and how to turn it on.

What OnRamp AI Does

OnRamp AI lives in the chat panel on the customer portal. It greets your customers when they arrive, surfaces what's changed, and tells them exactly what to work on next — and why. When they have questions, they can ask in plain language and get instant answers without hunting through their task list or adding a comment to the task.

It's a guide, not a gatekeeper. OnRamp AI points customers toward their work and keeps them oriented throughout their journey.

The First Visit

The first time a customer visits a project, the chat panel opens with a welcome message that introduces OnRamp AI and directs them to their first actionable task by name.

The message is specific to what they can actually do. If their task isn't ready yet — because a setup step is still in progress or a teammate hasn't completed a dependency — OnRamp AI explains that clearly instead of sending them somewhere they can't act.

The message includes a direct link to the task so customers can get there in one click. Alongside it, suggestion chips give them quick ways to ask follow-up questions — like Who needs to act first? — without typing.

Return Visits

When a customer comes back, OnRamp AI picks up where things left off. The return message leads with what matters most right now, citing:

• Overdue task — names the task and how many days it's past due
• Newly unblocked task — flags a task that was waiting and is now ready
• New assignment — surfaces a task that's been added since their last visit
• New comment — points them to a task with recent activity
• Progress update — summarizes what's been completed by the team
• Next task — if nothing else has changed, recommends where to pick back up

The response includes direct links to what needs attention, and suggestion chips update to match — so a return visit with an overdue task looks and feels different from one where everything is on track.

Suggestion Chips on Task Pages

When a customer opens a specific task, OnRamp AI surfaces chips relevant to that task:

• What does this task need from me? — plain-language breakdown of what's being asked
• What happens after this? — names the next task that will unlock once this one is done
• Why can't I start this yet? — explains what's blocking the task, whether that's a vendor step or a teammate's prerequisite, without exposing internal details your customer shouldn't see

Suggestion chips are always specific to the task the customer is looking at. They update automatically as the customer navigates.

Q&A

Customers can type questions directly into the chat panel at any time. OnRamp AI can answer questions like:

• What tasks are assigned to me?
• What's overdue?
• Who is on my team?
• What is this task asking me to do?

If OnRamp AI doesn't have enough information to answer something — like a vendor-specific process question or a commitment your team made outside the platform — it will say so clearly and suggest posting a comment or reaching out directly. It won't guess.

The Chat Panel

The chat panel can be positioned to fit your customers' workflow:

• Floating — hovers over the portal, accessible from anywhere
• Docked left or right — pinned to the side of the screen so it doesn't overlap task or comment content

Customers can move it themselves, and their preference is saved across sessions.

Every response includes a thumbs up / thumbs down control. Feedback is captured per message and helps OnRamp improve the experience over time.

Turning It On

OnRamp AI on the customer portal is controlled at the playbook and project level.

Default behavior:

• Existing projects and playbooks default to off
• New playbooks default to on
• New projects inherit the setting from the playbook they're created from

What OnRamp AI Won't Do

OnRamp AI is read-only and advisory. It will not:

• Surface internal tasks or vendor-only details that aren't visible in the portal
• Make up answers when it doesn't have the information
• Take any action without the customer initiating it

Frequently Asked Questions

Can I give OnRamp AI specific instructions for my project or process?
Not yet — but the ability to provide OnRamp AI with project-specific guidance (help documentation, support links, process notes) is coming. This will allow it to give more tailored answers for your customers' most common questions.

Why isn't OnRamp AI showing up when I preview the portal?
The portal preview tab in a project doesn't load the OnRamp AI experience. To see exactly what your customers will see, impersonate a customer user instead.

Will customers know they're talking to AI?
Yes. OnRamp AI introduces itself by name on the first visit and is clearly labeled in the chat panel.

Can I turn it off for specific projects without affecting others?
Yes. The project-level toggle lets you enable or disable it per project, independently of other projects in your org.

What if a customer asks something OnRamp AI can't answer?
It acknowledges the gap, tells the customer it doesn't have that information, and suggests a next step — typically posting a comment on the relevant task to reach your team directly.

Next Steps

Once OnRamp AI is enabled on a project, your customers are ready to use it on their next login.

To make the most of it:

• Make sure tasks have clear names and descriptions — OnRamp AI references them directly in its messages
• Attach resources to the project where they're relevant — these can surface as links in OnRamp AI's responses when customers ask questions about your project
• Keep task assignments current — OnRamp AI tailors its recommendations to who is assigned to what