The next release can deepen evidence for one dangerous agent action or broaden evaluation coverage across all agents.
Security reviewers need a defensible record of suspected misuse, while product adoption depends on covering more of the agent lifecycle.
Who you’d be doing this for
“My security lead asked why the agent did that, and I still can’t replay the whole decision.”
Rocio Cardoso · AI Application Engineer
Operates a customer-facing support agent that can trigger account actions.
What is at stake
A design partner cannot reconstruct a suspected prompt-injection action from its trace and evaluations. You must weigh a defensible investigation path against broader evaluation coverage in the only committed release slot.
Why it isn’t already fixed
Every obvious fix costs something else. That’s the part you’d have to decide.
- security evidence vs evaluation breadth
- release certainty vs investigation depth
- automated signals vs human verification
- design-partner urgency vs platform coherence
Why Datadog
Agent Observability connects agent traces and evaluations with production telemetry, making the evidence behind an agent action part of how customers operate AI in production.
Written with these in mind
Not your kind of problem? 40 more at Datadog, or browse every organization.
This is the setup. The work is inside.
Running it puts you in the room: the full situation and its constraints, stakeholders who push back in their own words, and the decisions that are yours to make. What you produce becomes a Day One Plan — work you can show someone instead of describing.