Ask real questions of real data, right inside Claude.
Agami connects your AI assistant to a governed dataset so you get trusted answers in plain English. This demo runs on real, publicly available data. Bring a question, get an answer, a chart, and the SQL behind it.
What is Agami
The trust layer between your AI tools and your data.
Agami lets anyone in your business ask a question in plain language and get an answer they can trust, from the AI tools they already use. Your data team owns the semantic layer underneath: the definitions, the joins, and the rules for what a good answer looks like. Agami turns that into governed answers, and shows the query behind every one, so nothing is a black box.
metric arr = SUM(mrr) × 12
grain account · fx usd
filters account_is_deleted = FALSE
account_is_deactivated = FALSE
account_is_test = FALSE
segments region · product_line · tier
vetted_by John Smith · 2026-03-15
→ Q2 ARR pacing +14% above plan.
NA $42.8M · EMEA $31.2M · APAC $18.6M
What's in the demo
California schools, end to end.
The demo is loaded with public California education data you can explore conversationally:
- K-12 (CDE). Every public school, census-day enrollment by race, gender and English-learner status, and free-or-reduced-price-meal eligibility, for the 2024-25 year.
- University of California. Freshman applications, admits and enrollees by source high school, campus, year and ethnicity, across all 9 UC undergraduate campuses, going back to 1994.
- Linked together. The two connect, so you can trace students from a high school all the way to UC admission.
Try asking
More datasets are coming. Want your own data in here? Talk to us.
How the data was built
Public sources, checked against the official numbers.
Every number traces back to a public source, and we validate it against the source of record before it goes into the demo. For the current dataset, that means public government and university publications: loaded, modeled, and reconciled to the publishers' own official totals. For example, exam pass rates, graduation rates, and enrollment counts all match the published figures. It is the same modeling and validation discipline Agami brings to your own data.
Spotted a number that looks wrong? Email us at contact@agami.ai and we will look into it. Public data has quirks, and we would rather hear about them.
Get started
Three steps to your first answer.
Add Agami to your assistant
In Claude, open Connectors, choose Add custom connector, name it Agami, and paste the URL https://demo.agami.ai/mcp. Click Add. It works the same way in ChatGPT and other MCP-capable assistants.
Open Claude connectors →New to this? See Anthropic's guide for Claude or OpenAI's guide for ChatGPT. Using Claude Code? Follow the Claude Code MCP docs.
Connect and sign in
Click Connect. Claude sends you to the Agami sign-in page, where you sign in with Google or Microsoft and pick the demo dataset. You are returned to Claude, now connected.
Ask away
Ask a question in plain English, for example "use Agami to show me the top 10 high schools in the Bay Area by test scores and UCB and UCLA admissions." Agami returns the answer, a chart, and the SQL it ran, so you can always check the work.
For richer charts and analysis, you can also add a data-exploration plugin, such as Anthropic's data plugin for Claude.
A note on AI and accuracy.
Agami uses AI to turn your questions into queries, and AI can make mistakes. Always check important numbers before you rely on them. The SQL behind every answer is shown so you can verify it. This demo uses public data only, so please do not enter any confidential information. The demo is provided for evaluation only, as is, and with no warranty. By using it you agree to our Terms of Service and Privacy Policy.