# Action Inspector

Click into any single activity and see exactly what your AI employee did: the model call pipeline, every tool it used with inputs and outputs, cost broken down by category, a duration waterfall, and the plan it followed. Available from the Activity tab, from Inspect on any chat message, or from the sidebar timeline.

Click any item in the activity feed and the execution inspector shows the full story. What was the original request. What the employee planned. Which tools it called, in what order, with what inputs. What each tool returned. What the employee decided based on those results. And what it produced as final output. Every step is visible.

The inspector shows the actual prompts sent to the AI model and the exact responses received. If you want to understand why your employee chose to send an email to the VP of Sales instead of the Marketing Director, the inspector shows the reasoning chain. If a tool call failed and the employee retried with different parameters, you see both attempts.

This level of transparency is what separates a trustworthy AI employee from a black box. When results are good, the inspector confirms the employee followed the right process. When results are bad, the inspector pinpoints exactly where things went wrong. You debug AI work the same way you would debug a human employee's decision: by understanding their thought process.

## See Every Step Your AI Agent Took, and Why

The Execution Inspector gives you a complete, step-by-step trace of every agent run. For each step, you see the tool that was called, the exact inputs passed to it, the output returned, the reasoning the agent used to decide on that step, and the time it took. Nothing is hidden.

This level of transparency is what separates trustworthy autonomous agents from black boxes. When an agent produces a result you did not expect, the Inspector shows you the exact decision path that led there. When an agent performs brilliantly, you can understand the strategy it used and replicate it across other employees.

## Debug Complex Agentic Workflows Without Guesswork

Debugging AI agents is fundamentally different from debugging traditional software. The failure point is rarely a code error. It is usually a reasoning choice: the agent misinterpreted context, selected the wrong tool, or made a plausible-sounding decision that led it off course. The Inspector gives you the data to diagnose these issues precisely.

You can step through the execution chronologically, jump to specific tool calls, expand any step to see the full input and output payload, and compare the agent's stated reasoning against what it actually did. This makes it possible to identify prompt issues, tool configuration problems, or unexpected edge cases in real-world data.

For multi-agent systems, the Inspector traces the full execution graph: which agent delegated to which, what inputs were passed at each delegation boundary, and what the receiving agent did with them. Debugging a 10-agent pipeline is as tractable as debugging a single agent.

## Execution Traces as a Compliance and QA Record

Beyond debugging, execution traces serve as a permanent record of what your AI employees did and how they did it. For regulated industries, this is not optional. Being able to show that an agent followed a defined process, did not access data it should not have, and produced output that can be traced back to specific inputs is a compliance requirement.

Traces are retained in full and are exportable in JSON format for integration with existing audit and QA systems. Each trace is linked to the activity feed entry and the original conversation, creating a complete chain of evidence from user input to agent output.

## How It Works

**Every agent run is broken into a complete step-by-step trace showing each reasoning step, tool call, input, and output in sequence.**

Select any task from the activity feed and open its execution trace. You see every step the AI agent took: what it thought, which tools it called, what parameters it passed, what each tool returned, and how it used that output to decide the next step. Latency and token usage are recorded per step so you can identify exactly where time and cost are spent.

The inspector is the primary debugging tool for complex agent behavior. If an agent reached a wrong conclusion, the trace shows you the exact point where the reasoning diverged. If a task is taking too long, the trace shows which step is the bottleneck. Engineers use it to optimize tool call sequences. Operations teams use it to verify the agent followed the correct workflow before approving a result.

## Use Cases

### Developer debugs a failed AI workflow step by step

The inspector shows every tool call, input, output, and decision the AI agent made, so the developer pinpoints exactly where things went wrong.

### Compliance team audits a completed agent run

Auditors open any past execution and walk through each step with full inputs and outputs, producing a verifiable record of what the AI employee did.

### Product team evaluates agent reasoning quality

The inspector exposes the agent's internal reasoning at each step, letting product teams assess quality and identify where prompts or tools need tuning.

### Support team investigates a customer-reported issue

When a customer reports an unexpected AI action, support opens the execution inspector and traces the exact steps the AI agent took.

## Comparison

| Before | After |
|---|---|
| Debugging AI agent failures means guessing from incomplete logs. | The inspector shows every step, every input, every output, in order. |
| Auditing agent behavior requires engineering to extract log data. | Compliance teams inspect any run directly, without developer help. |
| Agent reasoning is a black box with no visibility. | Every decision and tool call is exposed and inspectable. |
| Reproducing a bug requires rerunning the agent from scratch. | The inspector captures the full run, so reproduction is instant. |

## FAQ

### Can I see the reasoning behind each agent decision?

Yes. Each step in the trace includes the agent's internal reasoning (chain-of-thought) where available, showing why it chose a particular tool or approach. This is the primary tool for understanding unexpected agent behavior.

### How long are execution traces retained?

Execution traces are retained with your full activity history. There is no separate retention limit for traces. You can access any trace from any historical run through the Inspector or the API.

### Can I share a trace with my team or with support?

Yes. Each trace has a shareable link that gives read access to that specific execution. You can share it with team members for collaborative debugging or with Sistava support when investigating issues.

### Does the inspector work for multi-agent workflows?

Yes. In multi-agent runs, the Inspector shows the full execution graph across all participating agents, with clear delegation boundaries. You can view the trace for the entire workflow or drill into any individual agent's trace.

### How do I debug what steps my AI agent took to complete a task?

The step-by-step inspector shows a full execution trace for every task, including each tool call, input, output, and decision point. You can drill into any step to understand exactly what the agent did and why.

> The execution inspector showed us our AI agent was calling the same API three times per task. We fixed it in ten minutes and cut our costs by 40 percent.
>
> Sofia A., CTO · SaaS company

## Where Action Inspector fits

Action Inspector is part of How you see what they did.

A real-time activity feed shows who is working on what right now. The step-by-step inspector traces every tool call, decision, and reasoning chain. Cost tracking breaks down spend per message, per employee, and per team. Walk through your live 3D office to see your workforce at a glance.

- [How you see what they did](/en/features/observability): See everything. Miss nothing.

## Read the guide

- [Guide: Action Inspector](/en/guide/monitor/activity)

## More in Monitoring

- [Activity Timeline](/en/features/observability/activity_timeline): See every piece of work your AI employees do as it happens: chat replies, scheduled runs, tool calls, delegations, and finished deliverables, all logged as activities you can open and inspect. It runs automatically for every employee with no setup, and a live feed in the workspace sidebar shows actions as they occur so you never have to wonder what is happening right now.
- [See What Your Employee Is Told](/en/features/observability/prompt_inspection): Open any activity in the Activity Inspector and read the exact system prompt your AI employee received for that run: every instruction, every tool definition, and the manifest of which skills and duties were loaded and why. Nothing paraphrased, nothing summarized, the same text the model saw.
- [Credit Usage History](/en/features/observability/credit_usage_history): See exactly where your monthly credits go: split spend from direct employee requests and automation or system work, then inspect every charged system operation, including AI work, tools, Officer checks, knowledge, voice, meetings, video, and email. A burn chart shows the trend over time, and a separate grants table lists every bonus credit you received from promos, referrals, or purchases, with source and date.
- [Spending Limit](/en/features/observability/spend_limit): Set a ceiling on how many credits your workforce can spend per hour, day, week, or month, so a runaway task or a busy day never turns into a surprise bill.
- [Team Dashboards](/en/features/observability/team_dashboards): Every team gets its own dashboard the moment you open it: team health, the current sprint, work in flight, momentum, objectives, and KPI progress, all built from data the platform already has. On the Custom plan, connect data sources and the same dashboard adds Revenue & Pipeline, Demand & Conversion, and Reach & Engagement cards for that team.
- [Data Export](/en/features/observability/data_export): Every workspace surface that shows your history, chat threads, the task board, activity logs, and Drive files, has its own download control that exports the current, filtered view as Markdown, CSV, or JSON. Use it to pull records for an audit, share a run outside the app, or keep an offline copy of what an employee produced.
- [Review the Quality of Every Deliverable](/en/features/observability/quality_evaluation): After an employee does real work, a quality reviewer scores each deliverable against a standard and shows the verdict in the activity inspector. Everyday chat is left alone, so you only grade real output.
- [3D Office View](/en/features/observability/office_3d): A workforce of AI employees is easy to lose track of as a list of chat threads. The 3D Office View turns it into a place: every employee is a character at a desk, hot-desk, or the lounge, moving between them live as their status changes, so you can see your whole company working at a glance instead of clicking through tabs.
- [In-App Notifications](/en/features/observability/in_app_notifications): See the moment an Employee finishes work, needs a decision from you, or your credit balance runs low, through a bell icon in your workspace that always fires. Reply straight from the notification drawer for Employee updates and comment threads, and control which events also reach your email through a short list of categories.
- [Reply to Notifications](/en/features/observability/notification_reply): Reply straight from the notification drawer, whether it is an update from an Employee or a comment on your work, and your response goes back with the full context of what it is replying to. No need to open the chat or dig up the artifact first.
- [Email Notifications](/en/features/observability/email_notifications): Sistava emails you when an employee finishes work, needs your approval, hits a blocker, or your credit balance runs low, so nothing important gets stuck waiting for you to open the app. You choose which categories of email you want; account security and billing mail stay on because they carry information you need regardless of your other preferences.
- [Email Digest Notifications](/en/features/observability/email_digest_notifications): A rolled-up email that summarizes what your AI employees did over a set period, instead of a separate message for every event. Today, notification email in Sistava sends per event and per category (employee updates, task completion, credit alerts, and so on); a digest option that batches those into one scheduled email is planned but not built yet.

## Explore

- [Every feature](/en/features)
- [Hire an AI employee](/en/market)
- [Pricing](/en/pricing)