Context engine
Reads what your team reads.
Point Aurix at Slack, GitHub, Linear, email, or a folder. It builds a private index in minutes, no fine-tuning, no leaked data.
New · Aurix v2 agents are live
AI agents that
quietly
do the boring work.
Aurix connects to your stack, watches for patterns, and ships the follow-up, the update, the summary — so you can get back to the actual job.
- 1,400+Built for private automation workflows
- 1.1MEvent-driven agents
- 93.98%EU-hosted infrastructure
Running quietly inside
- Linear
- Notion
- Figma
- Stripe
- Vercel
- Raycast
Product
Small agents. Big relief.
Not another chatbot. Aurix deploys focused agents that read context, act with permission, and log everything.
Triggers
Fires on the thing that matters.
Webhook, schedule, mention, label, or "just watch that channel."
Safe actions
Ask-before-ship by default.
Approvals via Slack DM. Full rollback on every write.
Observability
Every thought, logged.
See the full trace — inputs, model calls, tool calls, outputs. Replay any run. Diff any change.
01parse(email)12ms
02classify(intent)148ms
03fetch(stripe.charges)402ms
04draft(reply)1.1s
05await(approval)—
Tool library
100+ connectors, or bring your own.
Pre-built for Gmail, GCal, HubSpot, Notion, Postgres, S3. Or drop in any OpenAPI spec and Aurix wires it.
GmailNotionPostgresHubSpotSupabaseGCalSnowflake+94 more
How it works
Ship your first agent before lunch.
-
01
Connect your stack
OAuth into the tools your team uses. Aurix indexes context privately — no training, no leaving your org.
-
02
Describe the boring job
"When a Stripe refund over $200 comes in, draft the apology, tag the CX lead, link the invoice." Plain English.
-
03
Watch it work, approve writes
First runs land in your inbox. Approve inline. Aurix learns your voice, gets faster, asks less.
Who uses it
Operators that ship quietly.
“We replaced a weekly 4-hour ops meeting with one Aurix agent. It summarizes, assigns, and pings only when something is genuinely off.”
“Our support team ships replies in 90 seconds instead of 14 minutes. Aurix drafts, we approve, customers are happier.”
“The approval-first model is the thing. I actually trust an agent to touch Stripe because I can veto anything in one tap.”
Pricing
Pay for runs, not seats.
Every plan includes unlimited teammates, full logs, and approval workflows.
Starter
$0/ forever
For solo tinkerers shipping their first agent.
- 500 agent runs / month
- 3 connectors
- Community support
- 7-day run history
Team
$39/ month, usage-based after
For growing teams that want approvals, SSO, and real limits.
- 10,000 agent runs / month
- All 100+ connectors
- Approval workflows
- 90-day run history + replay
- Priority support
Enterprise
Customannual
For regulated teams with VPC, audit, and custom retention.
- Unlimited runs
- VPC + self-hosted option
- SOC 2 · HIPAA · DPA
- Dedicated solutions engineer
FAQ
Straight answers.
Does Aurix train on our data?
No. Your content stays in your workspace. We use it to respond to agent runs, never to train base models.
Which LLMs power the agents?
Aurix routes across frontier models (Anthropic, OpenAI, Google) and can be pinned to a single provider on Team and Enterprise plans.
Can I self-host?
Yes. Enterprise customers can deploy Aurix in their own VPC with customer-managed KMS keys.
What happens if an agent does something wrong?
Every write action has a rollback. Runs are audited, diffable, and replayable. You can also gate any action behind a human approval.
How long does setup take?
About fifteen minutes for a first agent. Most teams ship three in the first week.
Do you offer startup discounts?
Yes. Seed-stage startups get the Team plan free for 12 months. Email founders@srv-claude.xyz.