Use this page to learn the few ideas you need to ask the agent good questions and get useful answers.

Alloy organises your data around a few concepts: missions, reports, scenarios, and the agent that works on them.

---

Missions

A mission is a single data collection session: one flight, one survey run, one field test. When you upload log files to Alloy, they become a mission.

What goes in

Every mission starts with your raw inputs:

• Data files: sensor logs, telemetry recordings, camera feeds, and other data from your autonomous system. Alloy supports MCAP and other common robotics log formats; custom formats are available upon request.
• Metadata: attributes that describe the session and device: software version, firmware ID, serial number, test configuration, operator, and any custom fields specific to your setup
• Description: your context about what happened: goals, conditions, observations, a ticket link. The agent uses it when it analyses the mission.

What comes out

Alloy processes these inputs into a single mission record containing:

• Reports: written by your report automations when the mission arrives (for example the First-look report), plus any report you ask the agent for
• Plots, maps and 3D trajectories: when the data supports them
• Structured metadata: extracted and user-provided attributes that power filtering and comparison
• Comments: notes from your team, which the agent also reads

How it connects

Missions are the top-level unit of organisation. Everything in Alloy connects back to missions:

• Search finds missions, reports, scenarios and chats by name and content, and hands deeper questions to the agent
• Reports analyse one or many missions
• Scenarios check your missions for a pattern you describe

---

Moments

Often the question is not about a whole mission but about one moment in it: the second the robot stopped, the frame where the obstacle appeared. The agent finds moments inside your recordings. It searches log messages by meaning, images by what is visible, and signals with SQL, then tells you the mission and the time.

Ask "find all moments where the robot detected an obstacle" or "show me where GPS signal was lost". The agent answers with each match and a link that opens Replay at that time, so you can check it yourself.

The fastest way to ask is Search: open it with Ctrl+K (⌘K on Mac), type what you're looking for, and press Enter to ask the agent.

---

Reports

A report is a shareable document that the agent writes from your mission data: narrative, charts, metrics, maps, event timelines and downloadable files. A report can cover one mission or many.

Automatic reports

Report automations (called Recurring reports in the app) write a report once you set them up. They run on each new mission after processing, or on a schedule.

• On each new mission. New organisations start with a First-look report automation. It answers what happened during this mission? with the evidence behind it. You can ask the agent to change it.
• On a schedule. Set a report prompt to run daily, weekly, or fortnightly. Useful for fleet performance summaries, anomaly digests, or any analysis you want refreshed regularly.

Custom reports

Custom reports are reports you ask the agent for. They can span any scope you need:

• Longitudinal analysis: compare performance across 50 missions over the last month
• Cross-version comparison: "compare software v2 with v1 across my last 2 sprints"
• Cross-device comparison: how does robot A perform vs. robot B on the same route?
• Deep dives: focus on a single metric or event type across your entire history
• Trend analysis: track how battery degradation progresses over weeks of operation
• Targeted investigation: "show me every mission where the IMU readings were anomalous"

Describe what you want, and the agent searches your data, runs the analysis, and builds the report. You can iterate until the report says exactly what you need. Reports live on the Reports page and are shareable by URL with your team.

---

Scenarios

A scenario is a check you describe in plain language. Alloy runs it on your missions from the past 7 days by default (up to 90 days) and on each new one, and lists the matches. It finds what you describe, nothing more.

Start with something simple like "find GPS issues". The agent works with you to turn it into a precise definition before the scenario starts.

Examples:

• "Find every flight where battery voltage dropped below 11V"
• "Flag any mission where the vehicle exceeded a 30-degree roll angle"

Each match shows the mission, the time window, and the reason it matched. New missions are checked as they arrive.

---

The agent

The agent is the AI agent that works on your mission data, in the app, in Slack and in Jira. Use it to:

• Ask questions about your mission data
• Find specific events or patterns across missions
• Create and refine custom reports and report automations
• Set up scenarios
• Run analysis with SQL and Python
• Generate charts, maps, metrics, and timelines
• Open Replay at the moment it is talking about

The agent can search and query every mission your organisation has uploaded. Ask it for the missions, times and queries behind an answer, or to open Replay at the moment it cites, so you can check the answer yourself.

What the agent remembers

The agent builds up context on your team, in your workspace. Each kind has its own owner and scope:

• Notes. Tell the agent to "remember that..." and it keeps the note for later chats, for just you or for your whole organisation.
• Scripts and documents. Save your team's analysis scripts and reference documents in the Scripts library. Every user in the organisation shares them. The agent says when it uses a saved script and when it writes new code. If a script's language cannot run in Alloy, it makes an equivalent version and keeps your original beside it.
• Past cases. In support roles, each request becomes a case with its outcome, so the next similar issue starts from what worked.
• Reports and comments. Past reports and comments on missions stay available to the agent. When you refine a recurring report in chat, the change applies to all its future reports.

A script tells the agent what to compute. It does not tell it how your team reads the result. Add that as a note: "a heading error over 0.5° on a straight run means a mounting problem".

Support roles

Alloy can set the agent up to answer in your support Slack channels and Jira projects, with instructions for each place. It keeps a record of each request and its outcome. See Slack.