How the Query Engine Works in Vantage
Ask questions about your product in plain language. Get answers sourced from your actual data, with citations you can click through to verify. No guessing. No searching through ten tools to find the answer.
The core problem
Product teams generate enormous amounts of context across their tools. Slack conversations about feature direction. Amplitude dashboards showing user behavior. Figma files with design iterations. Linear tickets with engineering constraints. When someone asks "why does the checkout flow work this way?" or "what did we learn from the last onboarding experiment?", the answer exists somewhere in these tools. Finding it takes thirty minutes of searching across five platforms.
How it works
Ask in plain language
Type your question the way you would ask a colleague. "Why did we change the pricing page layout?" or "What was the conversion rate for the new signup flow?" or "Which features were requested most in the last quarter?" The query engine understands product context. You do not need to use special syntax or know which tool has the answer.
Search across all connected data
The query engine searches across every connected data source simultaneously. It does not just search one tool at a time. A single query might pull context from a Slack conversation, an Amplitude metric, a Figma design, and a Linear ticket to construct a complete answer.
Get cited answers
Every answer includes citations linking to the specific source data. You see exactly where each piece of information came from. Click a citation to open the original Slack message, the Amplitude chart, or the Figma frame. If the query engine cannot find a sourced answer, it tells you rather than guessing.
Use answers in your workflow
Query answers are not just for reading. You can pull cited data directly into a spec, use query results to inform new requirements, or share answers with stakeholders as a link. The query engine is a tool for working, not just a search bar.
Example queries and what you get back
“Why did we change the onboarding flow last quarter?”
Sources a Slack thread where the head of product discussed drop-off rates, an Amplitude funnel showing the specific stage with the highest abandonment, and the PRD that documented the change.
“What are the open requirements for the billing project?”
Returns a list of requirements from the billing spec that do not have completed tickets, with links to each requirement and its connected ticket status.
“Has anyone else worked on notification preferences?”
Finds three past projects that touched notification preferences, shows the decisions made in each, and flags any active projects that currently reference notification settings.
“What compliance standards apply to our user data handling?”
Lists the compliance checks that have been run against specs involving user data, shows which standards were flagged, and links to the specific requirements and their resolutions.
How this is different from search
Traditional search in Slack, Notion, or Jira returns documents and messages that match your keywords. You still have to read through the results, find the relevant parts, and synthesize the answer yourself. The query engine does this synthesis for you:
Traditional search
- Returns documents that match keywords
- Searches one tool at a time
- You synthesize the answer manually
- No citations or verification
Vantage query engine
- Returns synthesized answers to your question
- Searches all connected tools simultaneously
- Answer is pre-synthesized with context
- Every claim cited with clickable sources