How it works
What Gauss does for your team.
Universal integrations, teachable knowledge, independent routines, and an audit log that leaves nothing out. Slack is where you interact with Gauss — but the substrate is the same wherever you reach it.
Integrations
Connects to everything you already use.
Gauss talks to every tool your team already uses — Google Workspace, GitHub, Notion, Linear, Salesforce, HubSpot, or anything else that speaks Model Context Protocol (MCP) or exposes an OpenAPI spec. Every call runs under credentials you supply. Some are workspace-level, so the whole team shares them; others bind to a single teammate via OAuth, so each person only reaches their own data. Either way, the credentials are yours, not ours.
- MCP servers, OpenAPI specs, and pre-built connectors for the usual suspects
- Per-user OAuth so each teammate only reaches data they already have access to
- Workspace-level connections for tools the whole team should share
#cs-team-leads
S
Sarah
@gauss how many CS tickets came in yesterday?
G
GaussAPP
Checked Zendesk over the last 24h — 47 tickets, up 12% from Tuesday.
47
tickets
12
% up
3
SEV-2
Playbooks
Teach it knowledge and skills.
Playbooks capture the things your team knows how to do — incident triage, weekly digests, onboarding, post-mortems, quarterly reviews. Gauss picks the right one automatically when the situation matches, and responds normally the rest of the time. Every teammate contributes to what Gauss knows; the shared knowledge base gets stronger every time somebody writes one down.
- Anyone on the team can write and edit them
- Right playbook picked automatically when the situation matches
- Version-controlled through the dashboard — see who changed what
📄weekly-cs-digest.mdmarkdown
## When to use
Weekly, Monday 9am. When someone asks "what happened in CS last week".
## Steps
1. Pull ticket counts from Zendesk
2. Compare against previous week
3. Flag any SEV-2+ that are still open
4. Post to #cs-team-leads as bullets
Routines
It doesn't wait to be asked.
Gauss doesn't need to be prompted every time. Give it a schedule in natural language — “every Monday at 9, post the CS digest,” “every hour, check the alert queue and page someone if there's a P1” — and it runs on its own, using the same tools, playbooks, and credentials as a live conversation. Auto-pauses if a routine keeps failing so you find out from a ping, not from a silent gap in the digest.
- Natural-language schedules — no cron syntax to learn
- One-shot or recurring; five consecutive failures auto-pause the routine
- Every scheduled run leaves the same trail as an interactive one
+ New
09:00
14:30
yesterday
3d ago
Auditability
Verifiable AI, not black-box AI.
Every prompt, every tool call, every response, and the exact model that produced it — all recorded to an append-only audit log scoped to your workspace. Sensitive strings are stripped before write. Export by user, by time range, by tool. If you can't verify what an AI did, you can't trust it in production. We start from that assumption.
- Sensitive strings stripped before write; export the log any time
- Cost, latency, and model attribution on every action
+ New
09:00
14:30
yesterday
3d ago
Ready to see the pricing?
Every markup percentage is published. The calculator on the pricing page shows the math.