Agentic analytics for your databases

Ask once. Every source answers.

Connect PostgreSQL, MySQL, SQL Server, SQLite or a spreadsheet. Each source gets its own data agent that writes and runs read-only SQL; a lead agent joins their results into one plain-English answer.

Compare revenue by region against this year's targets
Read-only by design Up to 5 sources at once No SQL required
run · revenue_vs_target3 agents online
Agent topology PostgreSQLsales_db MySQLcrm targets.csvupload Data agentrun_sql · ok Data agentrun_sql · ok Data agentrun_sql · ok Lead agenthandoff × 3 · join AnswerNorth +12% vs target
Agents query their own source in parallel. Only the results travel to the lead agent; credentials never reach the model.
Connects to PostgreSQLMySQLSQL ServerSQLiteCSV & ExcelJSONParquet
One agent per source

Each database gets a specialist that knows its schema.

A data agent sees only its own tables, writes SQL in that source's dialect and runs it. When a query fails, it reads the error, fixes the SQL and tries again.

  • Several sources answered in parallel, then joined by a lead agent
  • Schema, sample values and your glossary used as context
Queries this run2 sources
sales_db · PostgreSQLfailed → retried
SELECT region, SUM(amount) FROM order
sales_db · PostgreSQL4 rows
SELECT region, SUM(amount) AS revenue FROM orders GROUP BY region;
targets.csv · DuckDB4 rows
SELECT region, target FROM targets;
Answers you can check

Every step is on the record.

The answer comes with the SQL behind it and a trace of each handoff and query, so you can verify the numbers before you act on them.

  • Open any query's result table with one click
  • Follow-up questions reuse earlier results
Agent trace4 steps · 6.2 s
  1. Lead agent hands off to sales_dbtask: revenue by region, this year
  2. Lead agent hands off to targets.csvtask: regional targets, this year
  3. Data agents run 3 queriesrun_sql × 3 · 1 retry
  4. North is 12% above target; South trails by 4%answer · 168 words
From rows to decisions

Charts, anomalies and forecasts on any result.

Turn a result into a chart, flag unusual values with statistical checks, or project the trend forward with ARIMA forecasting.

Revenue by monthforecast · 3 months
anomalyforecast JanJulOct
Safe by default

Agents can read your data. They cannot change it.

A read-only guard is built into the query tool itself, and external databases run each query inside a transaction that is always rolled back.

Read-only guardactive
run_sqlblocked
DELETE FROM customers;
run_sqlblocked
SELECT * FROM read_csv('/etc/passwd');
run_sqlallowed
SELECT COUNT(*) FROM customers;
In the workspace

Built for the team that has to explain the numbers.

Keep the answers that matter, share them, and get them delivered on a schedule.

Dashboards

Pin results, charts and forecasts to a live overview.

Share links

Read-only links for stakeholders, no login needed.

Scheduled reports

Daily, weekly or monthly answers by email.

Business glossary

Upload definitions and KPIs so agents use your terms.

Three steps

Connect, ask, decide.

  1. Connect

    Add one or more databases, or upload CSV, Excel or JSON files.

  2. Ask

    Type the question the way you'd ask an analyst.

  3. Decide

    Read the answer, check the SQL, chart it or forecast it.

Pricing

Start free. Upgrade when it pays for itself.

Prepaid monthly or yearly, cancel by not renewing. Pay by card or mobile wallet.

Free

0

10 requests to try it on your own data. Up to 2 saved connections.

Start free

Team

3,950 EGP / month

For departments: 1,000 questions a day, every feature, unlimited connections and tables.

Choose Team

Enterprise

Custom

Self-hosted inside your network, your own AI model, unlimited usage and direct support. Talk to us.

Contact sales

Put an agent on every database you own.

Start with 10 free requests, then pick a plan. Bring your own model: OpenAI, Gemini, Claude or a self-hosted server.