Introducing AI Sources: see what built every answer

Today we’re launching AI Sources — a direct way to see exactly what produced every answer in Basedash.

Ask a question as usual. Once the answer is complete, the work that produced it collapses into a compact “Analyzed for…” summary. Open Sources beneath the answer to inspect the context and queries behind the conclusion.

The answer stays front and center. The evidence stays one click away.

Every input, in context

An answer can draw from more than one kind of company knowledge. Sources gathers those inputs into one place:

• tables from your connected databases and warehouses
• definitions and SQL models
• charts the assistant referenced
• data sources and read-only MCP connections
• web pages used during research

Each item appears as a compact context chip, so you can see the shape of the analysis before opening any individual source.

This is not a generic citation list added after the fact. It is the context Basedash used while producing that specific answer.

From the answer to the exact query

Sources also shows every data query the assistant ran. Expand a query to inspect its SQL and preview the rows it returned.

That makes it easier to answer the questions that follow an important result: Which table did this number come from? What period did the query use? Which records drove the change? Was the calculation grouped the way we expected?

You can read the result first and move into the underlying analysis only when you need it.

Thinking gets out of the way

While Basedash is working, the chat continues to show the latest step. When the answer is finished, the long sequence of thinking and tool calls collapses into a simple elapsed-time summary.

The final answer no longer has to compete with its process. Sources keeps the useful provenance available without leaving a trail of expanded activity in the thread.

Changes stay distinct from evidence

Reading data and changing something are different kinds of work. Basedash now treats them differently in the conversation.

If the assistant runs a mutating query, creates a definition or automation, or updates AI context, that action appears as a clear card beneath the answer. Open the card to inspect its details.

This keeps provenance readable and makes any change easy to spot.

Ask first, verify immediately

AI analysis is more useful when the path from conclusion to evidence is short. With AI Sources, the same conversation holds both: the answer you needed and the context, rows, and SQL that built it.

1. Sign up for Basedash or log in
2. Ask a question in chat
3. Open Sources beneath the completed answer
4. Inspect the context, queries, SQL, and returned rows

Ask the question, then see the receipts.