Wake up to the root cause. Not to the alert.
Every service traced from day one, with nothing in your code. When one drifts, Odigos Autofocus captures what the code did while it happens. Your engineers and AI agents open the laptop to the answer.
Your AI can change what production captures.
Manual logs and conventional automatic traces capture what was chosen in advance. When an investigation needs more, dynamic instrumentation lets your agent collect new evidence from the running service. No instrumentation release.
Decided before the question.
- Manual instrumentation
- Developers choose which logs, spans and values to add to application code.
- Conventional auto-instrumentation
- Integrations provide ready-made traces for supported libraries and operations.
New question. Same capture points.AI can query the telemetry you collect. A missing function or value still needs new instrumentation.
Directed by the investigation.
Which decision rejected this order?
The trace shows checkout failed. Inventory succeeded; payment never ran.
The agent requests function capture through MCP
FulfillmentGate.assessPipelineCoherence The next failed request reveals the values
- region
- eu-west
- replica skew
- 47 ms
- threshold in code
- 10 ms
- decision
- rejected
Targeted capture on live traffic. Nothing redeployed.
47 ms is normal for eu-west. The code applies the same 10 ms limit to both regions, rejecting healthy checkouts.
From the OdiShop demo · Read the walkthroughGo deeper during the incident. Return to baseline after.
Your incident automation can ask Odigos for more evidence, too. In the OdiMall demo, a Dynatrace workflow added temporary capture when a problem opened and removed it when the problem closed.
New failing traces carried payloads and code context into the same backend. Detailed capture lasted for the investigation. This demo used configured automation, with no LLM in the loop.
See the Dynatrace + Odigos demonstration- Problem opens in Dynatrace
A workflow calls a webhook that requests temporary instrumentation through Odigos MCP.
- New traces carry the missing detail
HTTP payloads, messaging payloads and supported code attributes appear without restarting pods.
- Problem closes. Temporary rules come off.
The workflow removes the rules through MCP. Capture returns to its baseline.
Autofocus can start before an agent asks.
Odigos Autofocus deepens capture when a service’s behavior changes, collecting functions, arguments and return values on the affected path. Engineers and agents can begin investigating as that extra evidence arrives in the backend.
Odigos Central.One control plane for every fleet.
An investigation may target one function. Odigos Central gives your team one place to manage and govern capture across every Odigos installation.
View fleet health and versions, manage configuration across environments, and keep team access and capture rules under central control. The control plane runs in your infrastructure.
Explore Odigos CentralOrganization-wide controls
- Capture scope and approval
- Sensitive-value masking
- Authentication, RBAC and audit trail
Keep your stack. Give it deeper evidence.
- Automatic coverage to start
- Discover and instrument running services without adding instrumentation to application code. Follow requests across supported runtimes.
- Your existing telemetry backend
- Send traces, metrics and logs as OpenTelemetry. Keep the dashboards, alerts and investigation tools your team already uses.
- Your team controls the capture
- Set scope, permissions and masking centrally. Review capture activity and remove temporary instrumentation when it has answered the question.
Found the cause? Help your agent fix it.
Carry the function, values and failure path into a targeted code change. Then verify the deployed fix against fresh traffic.
Stop guessing. Ask production.
One agent. One service. See how Odigos helps answer a production question.