Why Provenance

Know what happened.
React before it matters.

Every system generates events. Most companies lose them in scattered logs, siloed databases, and disconnected tools. Provenance captures every interaction in one place — and makes it actionable.

Complete visibility

Every action, every resource, every system — one unified audit trail. No more piecing together logs from 5 different services.

Faster incident response

When something goes wrong, trace the exact sequence of events across services. Know what happened, when, and where — in seconds.

AI-ready data

Structured interaction data is exactly what AI models need. MCP integration means your AI assistants can query real operational data, not just documentation.

Proactive, not reactive

Subscriptions and alerts mean you know about events as they happen — not hours later. Celebrate milestones and catch problems in the same system.

Works at any scale

Schema-per-tenant isolation, configurable rate limits, queue-based notification delivery with retry logic. Built for production from day one.

Low integration effort

One POST request to record an interaction. That's it. No agents to install, no SDKs required, no schema migrations on your side. REST API, CLI, or MCP — pick your interface.

Real-World Impact

Not just tracking. A platform that works for you.

See how Provenance's core capabilities — MCP integration, metric alerts, and the subscription engine — solve real business problems.

AI & MCP

Supercharge AI with real-time context

When your AI agents have access to every interaction that's ever happened in your system, they stop guessing and start knowing. Provenance's MCP integration gives any AI assistant — Claude, Amazon Q, custom chatbots — a live connection to your operational data.

Customer support AI

A support agent asks "What happened to order #4521?" — the AI queries Provenance, sees the full trace (created → payment confirmed → shipped → delivery failed → refund initiated), and gives the customer a complete answer in seconds. No dashboard hopping, no ticket escalation.

Fraud detection

Feed interaction patterns into your fraud model. When a user account shows 50 password resets across 3 origins in 10 minutes, Provenance has that data ready. Your AI flags it, a subscription fires a webhook to your security service, and the account is locked — automatically.

Intelligent chatbots

Your customer-facing chatbot can answer "When was my last deployment?" or "How many orders did we process today?" by querying Provenance through MCP. No custom integrations — just point the bot at your provenance data.

Operational copilots

DevOps engineers ask their AI assistant "What changed in production in the last hour?" and get a structured answer pulled from real interaction data — not log files, not guesswork.

support-agent.txt
Customer: What happened to my order #4521?
 
AI queries Provenance via MCP...
 
Agent: Here's what I found:
 
Mar 15 14:30 CREATE ORDER via WEBAPP
Mar 15 14:31 CREATE PAYMENT via STRIPE
Mar 15 15:00 CREATE SHIPMENT via WAREHOUSE
Mar 16 09:15 UPDATE SHIPMENT via CARRIER
└─ {"status": "delivery_failed"}
Mar 16 10:00 CREATE REFUND via SUPPORT
 
Agent: Your order shipped on Mar 15 but
delivery failed. A refund was initiated
on Mar 16. It should arrive in 3-5 days.
Alerts & KPIs

Track milestones, not just failures

Most alerting tools only tell you when things break. Provenance metric alerts work both ways — fire when error rates spike, but also when your team hits a KPI target, when a campaign reaches a milestone, or when a deployment count crosses a threshold.

KPI milestone tracking

Set an alert for when your platform processes its 100,000th order. When the interaction count crosses the threshold, Provenance fires a notification to your Slack channel — your team celebrates in real time.

SLA monitoring

Alert when a resource type hasn't received any interactions in 30 minutes. If your payment processor stops sending webhooks, you know immediately — not when customers start complaining.

Campaign performance

Track signup interactions filtered by origin. When your marketing campaign drives 1,000 signups, the alert fires and sends an email to the marketing team with the exact timestamp and breakdown.

Deployment frequency

Track your DORA metrics. Alert when deployment frequency drops below your target, or celebrate when your team ships 50 deployments in a week.

alerts.json
// KPI milestone — celebrate wins
{
"name": "100K Orders Milestone",
"metric": "count",
"resourceType": "ORDER",
"action": "CREATE",
"condition": ">=",
"threshold": 100000,
"subscriber": "slack-celebrations",
"adapter": "adapter:slack"
}
 
// SLA breach — catch problems early
{
"name": "Payment Processor Silent",
"metric": "count",
"resourceType": "PAYMENT",
"condition": "<",
"threshold": 1,
"window": "30m",
"subscriber": "ops-pagerduty"
}
Subscriptions

React to anything, automatically

The subscription system is the backbone of Provenance's event-driven architecture. When an interaction is recorded, a database trigger instantly checks for matching subscriptions and enqueues notifications. No polling, no delays, no missed events.

How it works

You create a subscriber (who gets notified), then a subscription (what triggers it). A subscription matches on resource type + action. When a matching interaction is recorded, the database trigger fires, finds all matching subscriptions, and enqueues a notification for each one.

Welcome emails

Subscribe to USER + CREATE. When a new user registers, Provenance enqueues an email via the SendGrid adapter. Template variables like {{interaction.name}} and {{interaction.email}} are substituted automatically from the interaction metadata.

Webhook pipelines

Subscribe to DEPLOYMENT + CREATE with a webhook adapter. Every time a deployment is recorded, Provenance POSTs the full interaction payload to your CI/CD pipeline, Slack channel, or monitoring service.

Multi-subscriber fan-out

One interaction can trigger multiple subscriptions. An ORDER + CREATE event can simultaneously send a confirmation email to the customer, a Slack notification to the sales team, and a webhook to your analytics pipeline.

subscription-flow.txt
1. Interaction recorded
POST /api/interactions { "type": "ORDER", "action": "CREATE" }
 
2. Database trigger fires
→ Finds 3 matching subscriptions
 
3. Notifications enqueued
→ adapter:email → customer@example.com
→ adapter:slack → #sales-wins
→ adapter:webhook → analytics.internal/ingest
 
4. Queue processor delivers
✔ Email sent (220ms)
✔ Slack posted (180ms)
✔ Webhook delivered (95ms)
 
5. Audit trail logged
→ 3/3 delivered, 0 failed

Who It's For

Built for teams that ship.

Engineering

Trace distributed transactions, debug production issues, track deployment frequency and DORA metrics.

Product

Build dashboards to track feature adoption, user journeys, and conversion funnels — without writing SQL.

Operations

Monitor SLAs, get alerted on anomalies, process notification queues, and maintain complete audit trails.

Support

Give AI agents access to customer interaction history. Resolve tickets faster with full context, not guesswork.

Start tracking today.

Free tier includes 10,000 interactions per month, full API access, dashboard UI, CLI, and MCP integration.