ChatGPT can build a financial model in a conversation. The difficulty is what happens next: the result is a file, the next conversation starts from a new upload, and the figures have to be checked again. A plugin changes where the model lives. Instead of a file ChatGPT generates, it is a model ChatGPT opens, edits and leaves in place.
Layerz is an official ChatGPT plugin for that job. This page describes what it does, how to add it, what ChatGPT can and cannot change, and how it fits with other assistants.
What Is the Layerz Plugin for ChatGPT?
It is the connection between ChatGPT and a Layerz model. Layerz is a structured spreadsheet, ready for AI: each model keeps its structure (items, formulas, timelines, links between the P&L, cash flow and balance sheet) separate from its data (actuals, scenarios).
The plugin passed OpenAI's review and has been listed in the ChatGPT plugin directory since October 2026. The same listing appears in Codex, OpenAI's coding agent. Underneath, it is a remote MCP (Model Context Protocol) server at https://layerz.cc/mcp, the open standard assistants use to call external tools.
What Can ChatGPT Do With It?
The work of a modeller, on a model that persists:
- Create a model from a description, or from a template (three statements that tie, an assumptions section, a balance check).
- Read a model: its items, formulas, dependencies and computed values, and its FINANCE.md, the file of conventions (currency, units, sign, fiscal year, glossary) every session reads first (FINANCE.md).
- Change it: add an assumption, rewrite a formula, restructure a section. The change propagates through the dependencies instead of breaking them.
- Load actuals from Excel, CSV, FEC or DATEV files, or from Pennylane, Qonto and Stripe through native connectors, and map them to the model's lines.
- Build scenarios as variants of the base case, and compare them.
- Check it: run the model's validation (balance sheet ties, cash closes, no orphan lines) and report what fails.
- Export it to Excel with formulas intact, for a board, an investor or a client who never opens Layerz.
Asking "what depends on gross margin?" gets an answer from the dependency graph, not from a scan of cells.
How Do You Add It?
From the directory, in a minute:
- Open the Layerz listing in ChatGPT's plugin directory.
- Add it and sign in to Layerz (a free account is created if you do not have one).
- In a conversation, ask ChatGPT to list your models or to create one.
No developer mode, no URL or API key to paste. The plugin directory is open on every ChatGPT plan. On a Business workspace, your admin manages which plugins are allowed; on Enterprise and Edu, plugins stay off until an admin enables them.
What Stays Under Your Control?
Three things, by design.
The figures. ChatGPT does not do the arithmetic. A deterministic engine evaluates every formula. ChatGPT proposes the change and explains the result; the number you sign comes from the model.
The changes. Each tool declares to ChatGPT whether it only reads or changes a model, so ChatGPT handles writes as write actions, not as lookups. Every change is recorded in the model's version history, with a diff against the previous state and a restore (24 hours plus one pinned version on the free plan, 30 days on Team).
The data. ChatGPT sees the models of the account you signed in with, and only those. For a stricter boundary, outside ChatGPT, an API key can be bound to a single model and set to read-only.
Does It Work Outside ChatGPT?
Yes. The model is not tied to the assistant that built it.
- Codex: the same plugin, from the Codex directory. Useful to script imports or checks around a model.
- Claude: connects over the same MCP server as a custom connector. The multi-client setup is in Claude for Fractional CFOs.
- Any client that supports remote MCP: same URL, same sign-in.
- Excel: every model exports to
.xlsxwith formulas intact.
A model created in ChatGPT on Monday can be reviewed from Claude on Tuesday and sent as an Excel file on Wednesday. That is what we call the CFO harness: the model, its conventions and its history sit outside any one assistant.
When Is the Plugin Worth It?
When the model lives longer than a conversation. For a one-off question on an export, ChatGPT's data analysis on an uploaded file is enough. The plugin is worth adding when the model is updated every month, read by someone else, or signed: a budget, a cash forecast, a client model in a fractional practice, a deal model. The difference is the one described in Persistent Memory for a Financial Model: the next session starts where the last one stopped.
FAQ
Is there an official ChatGPT plugin for financial models? Yes. Layerz is listed in the ChatGPT plugin directory after OpenAI's review, and in Codex. It lets ChatGPT create, read, edit and export structured financial models.
Is the Layerz ChatGPT plugin free? Layerz is free for one person working alone, with unlimited models. Team is for several editors on the same models. Your ChatGPT plan is separate.
Can ChatGPT edit my Excel model through the plugin? It edits a Layerz model. Import the Excel file into Layerz, let ChatGPT work on the structured version, then export back to Excel with formulas intact.
Does ChatGPT compute the numbers? No. The Layerz engine evaluates every formula. ChatGPT reads the results and explains them.
What is the difference with developer mode? Developer mode lets you connect any remote MCP server yourself, on the plans where OpenAI offers it. A plugin from the directory has passed OpenAI's review and needs no configuration.