Build your agent context like a file system
nao turns your metrics, schemas, and tribal knowledge into a structured context tree. Deploy a chat UI so anyone on your team can run analytics on your data—accurately.
Private beta · whitelist required · getnao.click
Context tree
using context/metrics/revenue.sql · definitions.md
Q4 net revenue retention was 114%, up from 109% in Q3. Expansion from the enterprise tier drove the gain, partially offset by SMB churn.
Why nao
Stop prompting. Start engineering context.
Most analytics agents hallucinate because they lack structure. nao gives your agent a curated, file-system-like memory so answers are accurate, explainable, and repeatable.
Context as a file system
Organize metrics, schemas, and definitions into folders and files. Version them, review them, and reuse them across every question.
Verified definitions
Pin the canonical SQL for each metric. nao answers from your source of truth instead of guessing—so numbers stay consistent.
Deployable chat UI
Ship a clean chat interface to your whole team. Non-technical teammates ask in plain English and get trustworthy answers.
Composable & inheritable
Scope context by team, project, or dataset. Child contexts inherit parents, so guardrails apply everywhere automatically.
Grounded answers
Every response cites the exact context files it used. Trace the reasoning and audit the query behind any number.
Connect your stack
Point nao at Postgres, Snowflake, BigQuery, or dbt. It maps your warehouse into context you can curate.
Context engineering
Treat your agent's knowledge like a codebase
A prompt is a blob. A file system is structure. nao lets you organize, version, and review the context your agent reasons over—so answers improve like software, not by luck.
- Folders group context by team, dataset, or domain
- Files hold metric SQL, schema docs, and business rules
- Verified files become your single source of truth
- Child contexts inherit guardrails from their parents
context/ ├── metrics/ │ ├── revenue.sql ✓ verified │ ├── active_users.sql ✓ verified │ └── churn.md ├── schemas/ │ ├── orders.schema │ └── events.schema ├── definitions.md glossary └── guardrails.md # guardrails.md - Always exclude internal test accounts - Fiscal year starts in February - "Revenue" = recognized, not booked
How it works
From raw warehouse to trusted answers in three steps
Connect your data
Link your warehouse or database. nao introspects schemas and drafts a starter context tree for you to refine.
$ nao connect postgres://...› mapped 42 tables → context/schemas/
Engineer your context
Curate metrics, definitions, and guardrails as files. Promote the queries you trust to verified status.
context/metrics/revenue.sql ✓ verifiedcontext/definitions.md glossary
Deploy the chat UI
Publish a chat interface to your team. Everyone runs analytics on the same trusted context—no SQL required.
$ nao deploy› live at chat.yourco.nao.click
Pricing
Whitelist-only while we're in private beta
Plans are activated for whitelisted teams. Request access and we'll unlock checkout for your workspace.
Starter
For small teams exploring context engineering.
- 1 connected data source
- Up to 5 context files
- Shared chat UI for 3 seats
- Community support
Team
For data teams running analytics across the org.
- Unlimited data sources
- Unlimited context files & versions
- Verified metrics & guardrails
- Up to 50 seats on the chat UI
- SSO & audit log
Enterprise
For regulated teams with advanced needs.
- Self-hosted or VPC deployment
- Role-based context scoping
- Dedicated solutions engineer
- Custom SLAs & security review
FAQ
Frequently asked questions
Instead of relying on one giant prompt, you structure everything your agent needs to know—metric definitions, schemas, business rules—into discrete, versioned files. nao loads the right files for each question, the same way an engineer organizes a codebase.
Text-to-SQL tools translate questions into queries with no memory of your definitions. nao grounds every answer in your curated context tree and cites the exact files it used, so results are consistent and auditable.
nao connects to Postgres, Snowflake, BigQuery, and dbt during the beta, with more warehouses on the way. It introspects your schema to bootstrap a starter context tree.
We're in private beta in 2026 and onboarding teams gradually to keep quality high. Joining the whitelist reserves your spot—we email you when your workspace is ready to sign in and purchase.
Yes. Once you've engineered the context, you deploy a chat UI where anyone can ask questions in plain English and receive trustworthy, sourced answers.
nao supports SSO, audit logs, role-based context scoping, and self-hosted or VPC deployment on Enterprise. Your data stays in your warehouse—nao queries it, it doesn't copy it.
Give your team an analytics agent it can actually trust
Join the whitelist for the private beta. We'll email you when your nao workspace is ready to deploy.