Ask about your systems. Get answers with evidence.

Anyone on your team can ask TraceX about your data, your code and what changed, in plain English, and see exactly how it found the answer.

Plain English

Questions in, cited answers out.

TraceX decides where the answer lives, queries your databases read-only, reads the code, and writes the answer with a citation for every number and record.

• Each step shown as it happens: queried, searched, read, checked
• Tables and charts for counts and trends
• Follow-up questions keep the context

How many refunds are stuck, and since when?

Queried refunds · 34 rows
Read 3 commits in payments-api
Checked the answer against the evidence
34 refunds are stuck in pending, all created after migration 0142 shipped on Oct 21 12.

Recent changes, explained.

Ask what shipped and what it did. TraceX reads the commits and deploys, checks the logs around them, and tells you what moved.

• Commits and deploys from GitHub
• Errors and logs from Sentry, Datadog and CloudWatch
• A citation for every claim

What changed in checkout since Monday?

Read 5 commits in web-app
Searched Datadog · checkout
2 changes shipped: a new payment form on Tuesday 1, and a retry on card errors on Wednesday 2. Checkout errors are flat since then 3.

Checked before you see it

A second model reads every answer next to the evidence; anything it can't confirm is corrected or marked as not confirmed by the data.

Knows your schema

TraceX learns how your tables link, so questions that span several tables get the right join.

Files when you need them

Ask for a spreadsheet, CSV, PDF or document and download it; exported files are deleted after 24 hours.

Ask in Slack

Mention TraceX in Slack and get the same answer, with the same evidence, in the thread.

Read-only, always

Conversations run read-only queries only, against the sources your admin connected.

Measured accuracy

Keep test questions with known answers and run them against live data, to see how often TraceX is right.

Counts and trends, drawn for you.

When the answer is a trend, TraceX shows it as a chart, built from the same read-only query it cites.

• Bar and line charts from your data
• The query behind every chart
• Export the numbers as a spreadsheet

How many refunds got stuck each day this week?

Queried refunds · 34 rows
34 refunds got stuck since Oct 21 1.

Checked

Every answer checked before you see it.

A second model reads each answer next to the evidence. Claims the data supports are kept; anything it can't confirm is corrected or marked.

• Each claim checked against the query results
• Unsupported claims corrected, not hidden
• You see exactly what was checked

Answer check 3 claims

34 refunds are stuck in pending Matches the query: 34 rows

Confirmed

They were all created after Oct 21 Earliest created_at is Oct 21, 10:03

Confirmed

Every affected customer is in the EU 6 countries, 2 outside the EU

Corrected

Conversation questions.

Who can ask questions?

Anyone your admins invite. Admins choose which sources are connected, and everyone's questions run read-only against them.

Does TraceX keep my data?

No copy of your database is kept. An answer stores what was looked up and where, not the rows themselves, and exported files are deleted after 24 hours.

Do I need to know SQL?

No. Ask in plain English; TraceX writes the query and shows it to you with the answer.

Find the root cause. Ship the fix.

TraceX is free during the private beta. Connect a database and a repository and ask your first question in minutes.