Use case

AI customer support for SaaS products

SaaS support has a specific shape: a long tail of how-do-I questions answered in your docs, a steady stream of billing and plan questions that need account data, and a small number of real bugs that need an engineer. Most tools treat all three the same.

SupportAi trains on your documentation and changelog, calls your API for account and plan questions, and escalates suspected bugs to your team with the conversation, plan and browser details attached.

Last updated

50–70%
of how-to tickets deflected to docs answers
1 inbox
for web, in-app, email and Slack Connect
Zero
engineer interruptions for non-bugs

The problem

What saas support teams tell us before they switch on an agent:

  • Tickets that are answered word for word in the docs
  • Billing and plan-limit questions that require looking up the account
  • Support requests arriving in Slack Connect channels and community
  • Engineers pulled into support for things that are not bugs

Example conversation

A typical exchange once the agent is connected to your systems.

  1. How do I add a teammate to my workspace?
  2. Go to Settings → Members and click Invite. You are on the Team plan with 3 of 5 seats used, so you have room for two more. Here is the docs page with screenshots.
  3. Can I get SSO on this plan?
  4. SSO is included on the Business plan. I can put you in touch with the team about upgrading, or you can compare plans here. Would you like me to flag this to someone?

Actions the agent runs

Described in plain English, pointed at your API. No code.

get_account

Returns plan, seat usage, billing status and feature flags for the signed-in user.

create_ticket

Opens a bug ticket in your tracker with conversation transcript and environment details.

check_status

Reads your status page API to report incidents before customers ask.

Also included for saas teams
  • Website and document knowledge with automatic reindexing
  • Web, WhatsApp, Slack and email in one inbox
  • Escalation rules on sentiment, keywords and fallbacks
  • Lead capture with webhooks to your CRM
  • Analytics and a daily digest
  • UK/EU hosting with a DPA on request

Docs that answer back

Point the crawler at your documentation, changelog and blog. The agent answers how-to questions with the exact steps and a link to the page, and reindexes automatically when you publish. Procedures let you define multi-step troubleshooting flows so answers are consistent.

Account-aware answers

Pass a signed user identifier from your app to the widget and the agent can call your API for plan, seats, invoices and feature flags. Billing questions get real answers, and upsell moments are surfaced naturally with lead capture.

Support where your customers work

Install to Slack for Slack Connect channels and internal helpdesks. Give the agent an inbound email address. Embed it in-app and on the marketing site. One knowledge base, one inbox, one set of escalation rules.

Protect engineering time

When a conversation looks like a real bug, the agent gathers reproduction steps, environment and plan, then opens a ticket via an action and escalates to a human. Engineers see clean tickets, not raw chats.

FAQ

Frequently asked questions

Yes. Pass a signed identifier to the widget and actions can fetch that user's account data securely.

Keep exploring

Train it on your docs in ten minutes

Paste your docs URL and see how many of last week's tickets it would have answered.

No credit card · Live in minutes · Cancel anytime