Everything between your user's question and your team's inbox.
One widget for your users, one inbox for your team, and the help center, status page and monitoring that keep both of them out of trouble.
Support that feels like part of your app.
A launcher in the corner that opens Home, Chat, Tickets and Help. It looks like your product, not ours.
- Your brand. Colours, font, corner radius and logo, set once in the portal.
- Context comes along. Screenshots, attachments and the page the user was on are included with every issue.
- Users stay in the loop. They follow their tickets in the widget and get an email when you reply.
- Company accounts. A customer's admin can see every ticket their colleagues raised.
- Voice. Users can talk to support instead of typing. Beta
Answers grounded in your product, not the internet.
Sortd answers from what you've given it, shows where each answer came from, and hands off when it should.
- Your help center, and your codebase when you connect GitHub. Code is searched where it sits, never pasted into a vector index.
- Live account data. Connect your API through MCP and Sortd can look up a user's plan, order or status, scoped to that user. You approve each tool first.
- Your voice. Set the tone and custom instructions, the topics that always go to a human, and the words to never use.
- Knows its limits. When it isn't sure, it asks a clarifying question or opens a ticket with the full conversation attached.
Every ticket arrives sorted and titled.
Users don't pick the right category, and they shouldn't have to. Sortd reads the report and decides.
- Question, bug or feature request, and it corrects the type when the user picked the wrong one.
- Titles that describe the issue, written from what the user reported rather than whatever they said last.
- Duplicates spotted before they pile up.
- Feature requests scoped with the user: Sortd asks follow-up questions and drafts a scope they approve.
- Routed to the right person with optional AI auto-assignment.
A queue that only holds real work.
By the time a ticket reaches your team, it's sorted, summarised and ready to act on.
- List or board, saved views, bulk actions and CSV export.
- Reply from the portal; the user sees it in the widget and by email.
- Alerts where you work: Slack, Microsoft Teams, Discord, email and webhooks.
- Bring your history. Import past tickets from a spreadsheet, export or document.
- See how it's going: satisfaction ratings, a dashboard and a live wallboard for the office TV.
Start with drafts, not a blank page.
The knowledge base is what Sortd answers from, so it's built to be easy to fill and keep current.
- Generated from your repo. Sortd maps your screens and flows, drafts articles, and you approve each one.
- KB Studio. A block editor with Studio AI: write a how-to, plan a section, apply your house style to every article. Beta
- Flags what it can't prove. Any claim without a source is marked for you to check.
- Move in easily. Import from Intercom, Zendesk, uploads or a website URL.
- Find the gaps. See what readers searched for and couldn't find, ranked by demand.
One public home for docs, status and what's new.
Publish your knowledge base as a fast, searchable help site in your brand. It's built from the same articles the AI answers from.
- Your domain, your logo, and search that jumps straight to the right section.
- Ask AI on the public site, answering only from published articles. Beta
- Article feedback so you know which pages aren't helping.
- Status page, roadmap and changelog alongside your docs, each switched on separately.
Know it's down before your users tell you.
Monitors watch your services and the vendors you depend on. When one fails, your team is alerted and your status page updates on its own.
- Three kinds of check: HTTP endpoints, vendor status feeds, and heartbeats for cron jobs and workers that have no URL.
- Incidents open and close automatically, and each one sends a single alert, not a flood.
- Tell users in the widget with an incident banner, so they don't open tickets about a known outage.
- Monitors as code. Define them in your deploy with one idempotent API call.
Connect what broke to who it affected.
Your backend sends exceptions to Sortd. They're grouped by cause, not by line number, so one problem stays one entry.
- Server-side, one endpoint. A single authenticated POST from your backend.
- Known issues, acknowledged. When a user reports something you're already tracking, they hear that it's being worked on.
- Off until you turn it on, with daily caps you control.