Documentation Index
Fetch the complete documentation index at: https://docs.aient.ai/llms.txt
Use this file to discover all available pages before exploring further.
Questions
These are the questions engineering leaders, operators, and agents usually ask before connecting Aient. The short version: Aient sits on top of your telemetry, turns noisy signals into problems, and helps drive each problem toward a reviewed fix.1. What is Aient?
Aient is an AI operations layer for software teams. It ingests telemetry, detects when something goes wrong, and groups related signals into a problem. From there, it explains what changed and drives remediation through Slack, GitHub, webhooks, and agent tools.2. How can Aient know what goes wrong?
Aient learns from the signals your applications emit: traces, logs, exceptions, browser errors, release metadata, and source maps. OpenTelemetry provides the runtime context. Source maps and commit metadata connect minified browser frames back to source. Integrations such as GitHub and Slack add the collaboration context Aient needs to route and explain each issue.3. What signals does Aient ingest?
The public integration surface accepts OTLP HTTP traces and logs. The Aient SDKs configure the endpoints, auth headers, service names, and release metadata for Node.js and browser apps. Webhook events and the Slack, GitHub, and Linear integrations add lifecycle and collaboration context around each problem.4. How do I add Aient to my services?
Install@aient/otel in server services and @aient/otel-browser in browser apps. Set an Aient publishable environment key, and pass a stable serviceName plus release metadata. For browser apps, upload source maps from CI with @aient/sourcemaps using the same service name and commit SHA.
5. What happens after telemetry reaches Aient?
Aient groups related telemetry into a problem, enriches it with service and release context, triages likely impact, and routes the next action. Depending on your configuration, that action can be a Slack notification, a webhook, an investigation, a GitHub issue, or a remediation PR.6. What is a problem?
A problem is Aient’s durable record for one customer-impacting issue. It is not a single log line or one trace. It is the grouped lifecycle around a failure pattern: detection, triage, routing, remediation, verification, closure, and possible regression.7. How does Aient avoid noisy duplicate alerts?
Aient treats repeated signals as evidence for the same problem when they share the same failure pattern and context. The product model is problem-first rather than event-first, so users see one evolving issue instead of a stream of repeated alerts. Routing policies and cooldowns control when Aient notifies again.8. Can Aient fix problems automatically?
Aient can investigate and prepare remediation work, including GitHub pull requests, when the integrations and policies allow it. Treat it as a draft author, not an auto-merge system. Humans keep review and merge authority unless you build a stronger approval workflow around it.9. Will Aient change production by itself?
No. Aient can open PRs, create issues, send notifications, and call configured workflows, but production changes still depend on your review, CI, deployment, and approval process.10. What do I see in Slack?
Slack is a collaboration surface for problem updates. Aient can post concise problem milestones, keep discussion threaded, and accept user replies or mentions where the Slack integration is configured. The canonical problem state remains in Aient.11. What does GitHub add?
GitHub lets Aient connect failures to repositories, commits, source files, issues, and pull requests. That context is what turns “there was a spike” into “this release introduced this failure and here is a candidate fix.”12. Where do source maps fit?
Source maps make browser crashes actionable. Without them, production stack frames often point to minified bundles. Uploading source maps with the matching service name and commit lets Aient resolve frames back to original files and lines.13. What keys do I need?
Use publishable environment keys in application telemetry code. Use server-side API keys only in trusted automation such as CI source-map uploads. Never ship a server API key to a browser or mobile app.14. What data should I avoid sending?
Do not attach secrets, auth tokens, raw request bodies, or sensitive user-entered content to spans or logs. Keep telemetry useful but minimal: service names, release metadata, error context, timings, and safe attributes.15. How do webhooks work?
Aient webhooks deliver tenant-selectable lifecycle events as CloudEvents structured JSON. Consumers should verify signatures, process events idempotently with the CloudEvents (source, id) tuple, and return success only after durable acceptance.
16. What do I use the MCP for?
Use the Aient MCP when an AI assistant or coding agent needs live Aient context. The current MCP shape is best for reading telemetry and problem context inside tools such as Cursor or Claude Code. Treat write actions such as buying, installing, or changing integrations as a separate provisioning surface until those scopes exist.17. How does Aient work with monorepos and many services?
Give each runtime a stableserviceName, set release metadata on every deployment, and upload source maps with the same service and commit. Aient then has enough context to group failures by service and connect browser stacks to the right build.