Using AI Chat#
Overview#
AI Chat is an AI assistant built into the Roboto web app that helps you explore, search, and understand your robotics data through natural conversation. Instead of writing queries or navigating through multiple pages, you can ask questions in plain language and get answers with clickable links to your data.
AI Chat is available in the chat sidebar (robot icon) and can be expanded to fullscreen for a more immersive experience. You can use Skills and Agents to automate your analysis — and AI Chat can help you create both. This guide covers the chat experience in the web app; for the underlying concepts — threads, agents, goals, and skills — see Agents and Threads.
Note
AI Chat is available for premium organizations. If you encounter any issues, please reach out and we’ll address them promptly.
Getting started#
Opening AI Chat#
To open AI Chat, click the robot icon in the sidebar of the Roboto web app. The chat panel will slide in from the side. You can expand it to fullscreen by clicking the expand icon at the top of the chat panel.
Context awareness#
AI Chat automatically picks up what you’re currently viewing. For example, if you’re on a dataset page, AI Chat knows which dataset you’re looking at and can answer questions about it directly.
Context items appear in the input area. You can remove them if you’d like to ask a general question instead.
This context awareness means you can navigate to a dataset or file first, then open AI Chat to ask specific questions about that data — no need to specify IDs or names.
Example prompts#
When you start a new chat, AI Chat suggests example prompts based on your current context:
On a dataset page:
“Briefly summarize this dataset”
“Link me to interesting visualizations”
“What topics are available?”
“Are there events in this dataset?”
On a file page:
“Briefly summarize this file”
“What topics are available?”
“Open a plot for me”
General (no specific context):
“What datasets were created today?”
“What are some tags I can search for?”
What AI Chat can do#
Search your data#
Find datasets, files, topics, and events using natural language. AI Chat translates your questions into RoboQL queries behind the scenes.
“Find datasets from last week”
“Show me files tagged with ‘crash’”
“What are some tags I can search for?”
Explore datasets#
Get summaries, list files, topics, and events for a specific dataset. Navigate to a dataset first, or reference it by name.
“Briefly summarize this dataset”
“What files are in this dataset?”
“List all topics”
“Are there events in this dataset?”
Explore files#
List topics and events for a file, and get file metadata.
“Briefly summarize this file”
“What topics are available?”
“Show me the events in this file”
Analyze topic data#
AI Chat can analyze topic data to extract statistics and find patterns — useful for quick analysis without writing code.
“What’s the average battery voltage?”
“Show me the min and max values for this topic”
“Are there any anomalies in the accelerometer data?”
Answer questions about your data model#
Understand what data is available, how it’s organized, and what events have been tagged.
“What topics does this file contain?”
“How many datasets do I have?”
“Find datasets with GPS data”
Help you create an agent#
Turn a conversation worth repeating into a reusable agent. AI Chat drafts the template, then opens the agent editor pre-filled; nothing is saved until you review and save the draft.
“Turn this conversation into a reusable agent”
“Create an agent that triages a dataset with these labels”
Help you create a skill#
Capture a workflow from the conversation as a skill your team can reuse. AI Chat drafts the name, description, and procedure, then opens the skill editor pre-filled; nothing is saved until you review and save the draft.
“Save this workflow as a skill”
“Create a skill from the steps we just followed”
Security and privacy#
AI Chat is designed with security as a priority:
Permission-scoped: It cannot access anything beyond your existing permissions. Security is enforced at the system level, not by the AI.
Scoped writes: Search and analysis are read-only. AI Chat can modify data only by applying tags, writing dataset summaries, and creating events. Goals you declare for a turn perform those writes as the turn runs; separate in-chat actions ask for your confirmation by default. AI Chat cannot delete data.
AI-generated content: Responses are generated by an LLM and may occasionally be inaccurate. Always verify important information.
Organization boundary: Your data stays within your organization’s boundary.
Current limitations#
It cannot access data outside your organization.
Complex multi-step analyses may require multiple follow-up questions.
It works best when you provide context by navigating to a dataset or file before chatting.
It cannot directly visualize data, but it can link you to the visualizer where you can explore interactively.
Going further#
Assign the AI verifiable work — triage labels, dataset summaries, event creation — with Goals.
Capture your team’s procedures as reusable Skills the AI can apply automatically.
Bring tools from external services into your conversations with MCP Connections (see the Connect External Tools via MCP guide).
Turn repeated workflows into reusable agents.