In development · Early access open
nanti builds specialized AI that plugs into Odoo — to lift sales margins, plan production, and act on your operational data. Built on Odoo’s data model. Designed to run on infrastructure you control.
For companies running Odoo — Community or Enterprise
The gap
Your Odoo already holds everything — orders, customers, stock levels, BOMs, margins, payment behavior. But turning that record into decisions still runs on exports, spreadsheets, and gut feel. The data lives in the system; the thinking happens outside it.
nanti closes that gap: specialized intelligence that works directly on your operational data, inside the workflows your team already runs — so the system that holds the answers starts producing them.
The approach
Generic AI doesn’t know what a reordering rule is, why a work order slipped, or which customers quietly stretch their payment terms. nanti’s products are built natively on Odoo’s data model and operational logic — they understand your business the way Odoo records it.
What we’re building
All four are in active development — no screenshots until there’s something real to show. Early access opens product by product, and the list gets in first.
Margin on every order. Intelligent upsell, cross-sell, and dynamic pricing computed from each customer’s actual history — surfaced to your sales team inside Odoo Sales, at quote time.
Built to move gross margin per order · attach rate
Planning that looks ahead. Reads demand, capacity, BOMs, and lead times to propose what to make and when — and replans the moment reality changes on the shop floor.
Built to move on-time delivery · inventory turns
Any LLM, safely inside your ERP. Query, report, analyze, and act on Odoo from Claude, ChatGPT, or your own local models — through a hardened MCP server with permissions your admin controls.
Built to move hours spent on reporting · time-to-answer
MCP — Model Context Protocol, the open standard for connecting AI to business systems.
Your AI. Your hardware. Your data. The full nanti stack with local models, designed to run entirely inside your own infrastructure — for companies whose data can’t leave, by policy or by law.
Built to move zero data egress · full auditability
Why nanti
Built on the data model, not around it. nanti treats pickings, BOMs, payment terms, and reordering rules as first-class concepts — because it’s engineered from Odoo’s own schema and logic, down to the field level.
Your data stays yours. nanti is being built to deploy in your private cloud or fully on-prem with local models — and we don’t train on your data, ever. Security isn’t a pricing tier here; it’s the architecture we’re designing from.
Measured on numbers you already track. Every product is designed against an operational metric — margin per order, on-time delivery, hours saved. If it doesn’t move a number that matters, it doesn’t ship.
Why we exist
Odoo won by being open and complete: one system for the whole business, owned by the business that runs it. Thousands of companies now hold their entire operational truth — every order, every movement, every invoice — in a system they control.
We think the next advantage comes from putting real intelligence on top of that record without giving up that control. Most AI asks you to ship your data to someone else’s cloud and trust a black box. We’re building the opposite: specialized AI, native to Odoo, measured on operational results, designed to run where your data already lives.
nanti is in development now, built alongside working businesses on real numbers. You won’t find case studies on this page yet — the first ones are being earned, by design partners who wanted these outcomes early enough to shape them.
Design partners welcome — talk to us
Plain answers
nanti builds specialized AI products for businesses that run on the Odoo ERP — an AI sales assistant, an AI manufacturing planner, an advanced MCP server for Odoo, and a secure on-prem AI deployment. They are built natively on Odoo’s data model rather than bolted on from outside. All four are in active development; early access is open.
We are building against both editions on current Odoo versions. Exact version and edition support is confirmed per product as each one reaches early access.
That is the explicit design goal of Secure On-Prem AI: local models, with data and inference staying inside your perimeter. It is in development, and the reference architecture is being defined together with design partners — we describe it as a goal, not a shipped product.
No. We do not train models on your business data. That is a standing commitment, not a pricing tier.
Early access opens product by product, in list order. Design partners get the first deployments and help shape what we build — if you want the outcomes early enough to influence them, talk to us.
Pricing is announced with each product’s launch. Early-access members hear first.
Early access
Early access opens product by product, in list order. Design partners get the first deployments — and shape what we build.
One email when a product opens. No newsletters, no spam.