PARTNERS / OPENAI
AtScale has been named an OpenAI Select Partner in the OpenAI Partner Network. With the AtScale plugin for Codex, Codex connects to AtScale over MCP and asks for the metrics Finance has signed off on. You keep the same agent and the same speed, and the figure it returns matches the finance report.
THE CHALLENGE
A company-wide rollout turns a definition problem into a volume problem. Finance, supply chain and the regional teams each ask in their own words, each get a defensible answer, and the answers don’t match. The disagreement surfaces in a meeting, long after the data review.
WHAT YOU GET
Every seat reads from one metric layer. AtScale Ava checks which metric someone meant, ACE computes it, and every answer carries the same definition.
SEE IT IN CHATGPT
Someone in the field asks a plain question. AtScale Ava (Ask, Verify, Answer) finds the five inventory metrics in your model and asks which one they meant, so the planning team and the warehouse team work from the same number.
You ask ChatGPT
"How much inventory do we have right now?"
↓ ATSCALE AVA CHECKS THE MODEL
AtScale Ava
Your model has five inventory metrics. Which one did you mean?
1. On hand
2. Available to promise
3. Net inventory
4. Days of cover
5. Inventory value
The weekly planning review uses 2.↓ ATSCALE ACE COMPUTES IT
1.24M units
Available to promise · all distribution centers
Computed by AtScale ACE · definition: inv_atp · owner: Supply Chain · excludes allocated and quarantined stock
Example. Metric names, definitions and figures are illustrative.
WHAT IT DOES
An administrator connects ChatGPT Enterprise to AtScale’s Model Context Protocol (MCP) server once, for the whole workspace. From then on, everyone’s reading the same metrics.
AtScale ACE (the AI Computation Engine) computes each figure where the data already sits, from aggregates it maintains. A rollout to ten thousand seats doesn’t become ten thousand scans of the fact table. The bill stays flat.
Developers get the same definitions in Codex, so when a coding agent builds a dashboard, it’s using the revenue definition the CFO signs off on.
HOW IT CONNECTS
01
An administrator configures AtScale’s MCP server once, for everyone.
02
Access rules in the model follow the person asking.
03
Ava checks with the person when one word maps to more than one metric.
04
Metric usage shows up in one place, so you learn which definitions the company runs on.
MORE WAYS TO USE IT
The model decides who sees which numbers, for every seat.
ACE answers from maintained aggregates, so more people asking doesn’t mean more full scans.
Find out which metrics people ask for, and which ones go unused.
Update a definition once and every answer follows.
The number lives in the model, so nobody emails a file around.
Codex builds against the same metrics ChatGPT answers with.
PROOF
Case Studies
Case Studies
Case StudiesFAQ
ChatGPT Enterprise and Codex, both through AtScale’s MCP server. An administrator connects it once for the whole workspace.
As many as you’ve licensed. ACE (the AI Computation Engine) computes each metric from aggregates it maintains, so more seats don’t mean more full scans of your fact tables.
Ava checks which metric they meant. It lists the metrics that match, like the five inventory metrics in the example above, and nothing’s computed until the person picks one.
Yes. Each person signs in as themselves with OAuth, so the model’s access rules follow them into ChatGPT. The MCP tools are read-only.
Yes. Metric usage shows up in one place, so you can see which definitions people request most and which ones go unused.