kudhubKeep the data queryable
Ask a new question without preparing the same files again.
A home for the data behind your answers
Organize and understand
your data with AI.
Bring files together through your AI assistant. Keep prepared data for the next question, the next batch, or a shared view.
Free during private beta. Bring your own assistant.
Setup guides for ChatGPT, Claude, Codex and more →
Northstar / Order dataFictional example
BRING THE RECORDS
- orders.csv
- payments.csv
- shipments.csv
Three files. One question that needs all of them.
YOU, IN YOUR AI ASSISTANT
“Which paid orders still need shipping? Save the combined data so we can use it again.”
NEXT TIME
“Here's the next batch. Use the same rules.”
ILLUSTRATIVE ANSWER
5 orders. €12,200.
Fully paid, with no shipment recorded.
| Country | Orders | Value |
|---|---|---|
| Spain | 2 | €6,400 |
| France | 2 | €4,800 |
| Portugal | 1 | €1,000 |
What stays useful: source files, saved matching rules, and a queryable table.See the records ↗
Why stop at a dashboard?
Keep the data behind every answer.
A chart answers one question. Ask your assistant to save the prepared records as a database table, and the same work can answer the next one.
Connect the next source
Join saved records with another file when the question grows.
Ask for only the rows you need
Give your assistant a focused result instead of whole files. Smaller answers can use fewer tokens.
Recover the records too
Compare retained versions and repair selected data after a mistake.
Savings depend on the source, query, and assistant. History and recovery need retained workspace data.
Start with the records you have.
Bring CSV or Parquet files, or connect Git, S3, ClickHouse, and public URLs. Your assistant can match records across sources and save the result.
Explore supported sources →orders.csvpayments.csvshipments.csv
SAVED IN THE WORKSPACEOrders ready to inspectOrder · Paid · Shipped1007 · Yes · No1001 · Yes · Yes
Ask the next question in your assistant.
Keep the conversation in your own AI client. Ask it to inspect the saved data, explain a result, or run the preparation on a new batch.
You choose when to refresh sources and run work.
Find your assistant's setup guide →“Which of those orders are going to Spain?”
ILLUSTRATIVE ANSWER2 orders. €6,400.Orders 1007 and 1010, from the data you already prepared.
Same records. A more focused answer.
Return to old data. Repair mistakes.
Ask an earlier version a new question. Or restore the records an agent got wrong while keeping unrelated later changes.
Historical questions need retained data. Live sources do not automatically preserve old records.
Explore a historical question →Earlier recordsNew batchAgent change
INSPECT BEFORE RECOVERINGWhat changed?Compare the saved versions, then restore only the records you choose.See selective recovery ↗
What would you like
to understand?
Six connected stories. One fictional dataset, with checked results at every step.
Which paid orders still need shipping? →
Match three files. Find five orders worth €12,200.
Use the same definitions this month. →
Compare batches: two open orders become five.
Answer here. Give the team a view. →
One dataset for a conversation, a CSV, and a dashboard.
What if we changed this rule? →
Test a €5,000 threshold without changing the accepted data.
Ask last year's data a new question. →
Run the same question against two retained snapshots.
Undo the part the agent got wrong. →
Repair two countries. Keep the later correction.
An answer now.
A view to come back to.
Two current outputs from the same prepared data. Fictional sample results.
An answer in your assistant
ILLUSTRATIVE ANSWER
“Five paid orders still need shipping, worth €12,200.”
Spain accounts for €6,400, France €4,800, and Portugal €1,000.
Ask next: “Give me just the open orders as a CSV.”
Follow up in the same conversation.
Follow the conversation →A view for your team
OPERATIONS DASHBOARD · OPEN ORDERS
- Order value
- €12,200
- Orders
- 5
| Country | Orders | Value |
|---|---|---|
| Spain | 2 | €6,400 |
| France | 2 | €4,800 |
| Portugal | 1 | €1,000 |
Preview privately, then choose who can see it.
Explore the working dashboard →CSV export is available today. See more sample outputs →
Where it could go next.
Not available today. Planned directions may change; explorations are ideas for beta feedback.
A report in your inbox
“Email me the results when this report is ready.”
A PDF for your client
“Turn this analysis into a branded report.”
Scheduled updates
“Check every Monday. Alert me when open orders exceed €15,000.”
Use your data in another tool
Let another application read selected prepared tables.
More ways to take results with you
More file formats and delivery options beyond CSV.
Bring your assistant.
Start with a question.
Three steps to your first saved result.
Check assistant setup →Get an invitation
Join the waiting list, or redeem a code if you already have one.
Connect your assistant
Create an agent credential, then follow your client's setup guide.
Bring data and ask
Tell your assistant which files to use, what to answer, and what it may change.
Private beta · Invitation only
Bring your data.
Bring your questions.
Help shape how kudhub works for the things you need to understand.
Free during private beta. Prices will be published before anyone is charged.
Request beta access
Leave your email. We'll contact you when an invitation is available.
Already have an invitation? Redeem it →