AI analytics
BI for the AI you already use.
Newtown builds your entire analytics stack in minutes, connects it to your LLM, and models your data as you go.
Works with Claude, ChatGPT and any MCP client. Free to bring your own warehouse or dbt project.
The problem
Every business needs one place to go for the truth.
Until now, that meant a warehouse, a BI tool and a data team: months of setup before the first trusted number.
Data-backed answers
Real numbers behind every decision, not gut feel or last month's export.
One source of truth
Every source in one place, so the answer doesn't depend on which tool you opened.
Metrics everyone agrees on
Revenue, churn and CAC defined once, and answered fast for the whole team.
Why now
Business owners already ask their LLM. Vendor connectors get them halfway.
Shopify, Stripe and HubSpot connectors in Claude or ChatGPT answer "what were sales yesterday?" in seconds.
- A governed metric. Revenue the way your team defines it, every time.
- A trend over time. Full history, not whatever the API returns today.
- Across sources. Meta spend against Shopify revenue against Stripe churn.
The product
Your entire analytics stack in minutes, in your LLM of choice.
Set up in minutes
Connect your sources and we create the warehouse, pipelines and dbt project. Or connect the stack you already have.
All your data, in your LLM
Every source queryable together in Claude, ChatGPT or any MCP client: defined metrics, trends and charts.
Modeled as you go
Our agents turn your questions into models and metrics in your own GitHub. Correct a number once and it becomes the definition.
You don't have to know what to model up front. Ask questions, and the questions drive the modeling. You own the repo, the warehouse and the definitions; nothing is locked inside Newtown.
How it works
From sources to answers, modeled as you go.
Your sources
Shopify, Stripe, HubSpot, ad platforms, app databases
Your warehouse
BigQuery or Snowflake: yours, or one we run for you
Newtown
dbt project, metric layer, modeling agents, guardrails
Your LLM
Claude, ChatGPT or any MCP client
The loop. Every question is logged. Gaps become modeling work that the agents pick up overnight, and your corrections become the definition.
Governed by default
Answers you can report on, without the back-and-forth.
A vendor connector always answers. Newtown says no when the number would be unreliable, and tells you why.
Computed confidence
Every answer is banded from the query plan. The model can't talk its own confidence up.
Refusals with reasons
Money questions over unmodeled data are refused, with what's missing and what it can answer instead.
Join guard
A cross-table join is refused until the relationship has been measured against the data.
Gaps become backlog
Ad hoc queries are logged as modeling gaps, so invisible work becomes a visible queue.
Corrections stick
Tell it how you define a metric once and that becomes the governed definition for everyone.
Your keys, your data
We never touch your source credentials. Each workspace has its own identity, and you decide what our team and your LLM can access.
Track record
A decade of data team work, distilled into a product.
The Newtown team has built and run data stacks for consumer, media and health companies. The platform turns that work into something every company can use.
“Newtown connected our ad sales, podcast delivery, and billing data in one place. Suddenly there was more information available than we knew what to do with.”Forrest Cao, Sr Director of Finance · Dear Media
“I can answer questions about conversion, segmentation, and platform performance in minutes that used to take hours or weren't answerable at all.”Yacha Toueg, Head of Product · obé Fitness
“Newtown embraces AI as a force-multiplier in their workflow, ensuring our team stays at the cutting edge of engineering efficiency.”Kelly Burdine, VP Data, Product, and Engineering Operations · Wellthy
Pricing
Free to bring your own. Pay for the work.
You pay for the data team's labor, not for plumbing. Questions run on your own Claude or ChatGPT subscription, so your bill doesn't grow when your team asks more.
Everything you bring
- Connect your warehouses and raw data
- Connect your existing dbt project
- Add business context
- Governed answers in your LLM
Everything we do for you
- New models and model changes, built by our agents
- Managed connectors and a warehouse we run
- Review by Newtown's data team
- No per-question metering
Next steps
Start with a 14-day design partner pilot.
Connect
Sources, warehouse or dbt repo, plus your context or Looker.
Model
We model the metrics your team asks about most.
Use it
Your team asks in Claude or ChatGPT; corrections become definitions.
Questions first? hello@newtowndata.io