Most teams keep an agent from regressing by hardcoding checks into its logic, an if-statement here, a regex there. Every new rule then becomes a code change, a review, and a deploy, and the person who spots the problem in production is rarely the person who can ship the fix. Agent Control moves those rules out of the code and into one hub. Steer, block, and validate agent behavior in real time, with rules any team member can update without touching the codebase. In this walkthrough, we break a live multi-agent stock scoring app and fix it in under two minutes: A healthy run: the Growth Analyst builds the bull case, the Risk Analyst the bear case, converging on a 3/10 consensus score for SPY The failure: the Risk Analyst leaks a client's name, SSN, and email straight into the response, a live compliance incident The fix with no code patch and no redeploy, a steer-pii-ssn control enabled and saved in the Control UI The same query re-run: identical tool call and analyst reasoning, but the SSN is intercepted and redacted before it reaches the user Trace View showing which control triggered, on which agent, and the evaluator result ("action": "steer", "matched": true) A control library spanning pre- and post-execution stages, LLM and tool steps, and server- or SDK-side enforcement, all custom-defined Every control decision lands in the trace alongside the rest of your observability data. If a behavior can be defined, it can be governed, build once, apply everywhere. Take command of your agents with Agent Control today. Try Splunk Agent Observability: https://www.splunk.com/en_us/download/observability-cloud-free-edition.html Docs: https://agent-observability-docs.splunk.com/what-is-splunk-agent-observability 0:00 The question nobody answers before shipping: what stops it from regressing? 0:12 Why hardcoded if-statements and regex don't scale 0:24 What centralized, external controls need to do 0:36 Agent Control: steer, block and validate in real time 0:45 The app: two analyst agents scoring the S&P 500 1:03 The leak: client name, SSN and email in the output 1:11 The fix: enable a PII control in the UI, no redeploy 1:26 Same query, control now enforcing 1:35 SSN intercepted and redacted, clean response delivered 1:47 Trace view: which control fired, on which agent, what result 2:08 Any control, any agent, any project — build once, apply everywhere 2:25 Centralized governance for every agent

 91      0