Why Ridgepole
The boring foundation that holds up the house
In short
- The AI can't invent your numbers
- Custom connectors, not generic ones
- A data science team, included
- Software for the repeats
- Your data stays in your cloud
- Every number shows its source
- We reconcile identity, not just move data
- Each build starts ahead
What makes it different
What makes the build different
The AI can't invent your numbers
The AI picks references to approved figures. Our server writes the values, units, dates and coverage notes, and rejects any pick it can't support. The model never writes its own database queries.
Custom connectors, not generic ones
We build each connector from our library to fit your systems. Generic connectors read the surface. Ours follow how your data is actually shaped.
A data science team, included
Our harness keeps complex analysis on track: how the data runs through, the statistics behind it, and results that are shown and proven. Developed by math PhDs and data scientists from Frontier and Meta.
Software for the repeats
Tokens are about to get expensive. Repeatable steps run as software you own. Model reasoning goes where it adds value: interpretation and exceptions.
Your data stays in your cloud
Everything runs in a private AWS environment that we set up for you. Healthcare builds include HIPAA technical safeguards.
Every number shows its source
We can show your CFO where a figure came from and what data was missing. When something breaks, you can see it.
We reconcile identity, not just move data
Anyone can copy tables. The hard work is making 'Location 12' in accounting mean the same thing as 'Building B' in another system. We map it and keep it current.
Each build starts ahead
120 pieces of reusable infrastructure carry from one build to the next. You get everything built before you.
Compared with
The other ways to get there
Each one is a real option. Here is what to ask of each.
One person, and a bus factor of one
Hiring a data analyst
Compare our fee with the salary plus the cost to build and maintain the infrastructure that person needs. One hire is one person's capacity. We deliver the whole capability: connectors, reconciliation, sources and AI answers, kept running.
They show what you feed them
Tableau, Looker, or Power BI
BI tools connect shallowly and don't clean data. Typically only 10% of the available data gets used. We bring the rest to light and connect your BI to the clean layer underneath.
Ask what they reuse
A large consultancy
Ask about their timeline and what they bring from past work. We don't rebuild from scratch each time. That is where our speed comes from.
Ask what stops it inventing numbers
An AI startup
Ask what their AI reads, and what stops it from making up a figure. Ours picks server-owned figures by reference and shows the source date and coverage with every answer.
Every run pays for tokens again
Building it in-house with AI tools
A chat tool rebuilds your company context on every question. We turn the repeatable steps into software you run, which costs less than token bills that keep rising.
Team
A lean, senior team
Ridgepole was cofounded by Jamey McDowell, a former Meta software engineer and data scientist.
We're lean, and that is why we're fast and affordable. No big-company red tape, and hands-on help when you need it. We run local models that do the bulk of the work.
FAQ
Questions we get asked
What do we actually get?
A data science team, BI and AI in one. We connect your systems, make the numbers agree, and build the reports, workflows, tools and AI answers you need on top. We keep it running.
How long does it take?
We scope a first deliverable after we check access, data and definitions. Reuse makes us fast, but we don't promise a date before we look.
What does it cost?
A one-time build fee plus an ongoing maintenance fee, with the AWS setup included. In discovery we scope both and give you an all-in year-one estimate, with an ROI case on your numbers: what hiring for this would cost, what manual reporting costs today, and what faster, trustworthy answers are worth.
Can your AI invent our financial numbers?
The AI selects from approved figures, and our server writes the values with their source dates. That blocks invented numbers. We still verify which metric is selected and that the data upstream is right.
Which systems do you connect to?
We build a connector for each system you use, from our library. If a system has an API or an export, we can connect it. We confirm the details for each system during discovery.
We already have a BI tool. Do we still need this?
Keep it. BI is the visual layer. Underneath it is the data, and that is what we own. We connect your BI to clean, connected data so it shows more than the usual 10%.
Where does our data live?
In a private AWS environment that we set up for you. We collect the minimum we need and control who sees what.
Is access to our data audited?
Sensitive paths in the application enforce audit logging. We walk you through exactly what is covered during scoping.
What happens if you disappear?
Your infrastructure is written as code, with documented migrations, runbooks, and a handoff plan in the contract. Ask us about it. We have an answer.
What would you like to understand or get done?
A question your systems can't answer. Work your team repeats. A tool you wish existed. Tell us where you want to start.