Notion Rolls Out Custom Agent Insights for Full Visibility Into Agent Runs.

Notion is addressing one of the biggest pain points in AI agent deployment: knowing what your agents are actually doing. The company announced Custom Agent Insights this week, a new capability that provides complete visibility into agent activity during every run.

The update tackles a common problem for teams using autonomous AI agents. When agents handle workflows on their own, teams often lack insight into the decision-making process, making it difficult to debug issues, verify outputs, or understand why an agent took a particular action. Custom Agent Insights aims to eliminate that uncertainty.

The new insights feature provides runtime visibility into agent behavior. Teams can now see the full sequence of actions an agent takes, from initial trigger to final output. This includes which tools the agent accessed, what data it pulled, and how it processed information to complete a task.

For organizations running Custom Agents, this means better oversight of autonomous workflows. Rather than treating agents as black boxes, teams can inspect each step of execution. If an agent produces an unexpected result or fails to complete a workflow, the insights make it easier to identify where things went wrong.

The rollout comes as Notion continues building out its AI infrastructure. Earlier this year, the company launched its Developer Platform, giving developers tools to extend agent capabilities with custom workers and integrations.

Agent transparency has become critical as teams deploy AI for more consequential workflows. When agents handle customer data, generate content, or make operational decisions, teams need audit trails. Custom Agent Insights provides that paper trail, showing exactly what happened during each run.

This matters particularly for regulated industries or teams with compliance requirements. Being able to trace an agent's actions from start to finish helps meet documentation standards and provides evidence of proper handling.

The feature also supports iterative improvement. Teams can review agent runs to identify patterns, spot inefficiencies, or find opportunities to optimize workflows. Rather than guessing why an agent behaves a certain way, they can see the actual execution path.

Notion has been moving aggressively into the AI agent space. Custom Agents, currently in beta, allow teams to build autonomous workflows that handle repetitive tasks without human intervention. Recent case studies show companies like Vercel using these agents to accelerate shipping cycles.

The Developer Platform extends agent capabilities further, letting teams connect agents to external data sources, build custom tools, and sync information across systems. Custom Agent Insights fits into this ecosystem by making agent behavior observable and debuggable.

The feature is rolling out now to teams with access to Custom Agents. Notion has not specified whether insights will be available across all plan tiers or limited to enterprise customers.