How to use ChatGPT with SAP Business One
Short answer
There are three ways: paste or upload exported data, give a custom GPT or ChatGPT app limited access to a small API in front of the Service Layer, or skip the public service and run a private model. The first is quick but manual, the second needs a hardened public endpoint, and the third suits regulated or price-sensitive data.
Many Business One users already use ChatGPT on the side. Someone exports an item list from a query, pastes it into a chat, and asks for a summary or a cleaned-up description. It works, and for non-sensitive data it is perfectly reasonable.
The next step is wanting ChatGPT to answer questions about live data: what is the status of order 4521, which customers are over their credit limit. That is where the real design decisions begin, because the model is hosted outside your network while your Business One server probably is not exposed to the internet.
This page lays out the realistic options, what a custom GPT action needs from your side, what data controls to check, and when it is cleaner to keep the model in-house.
What teams use it for
Clean up item descriptions in bulk
Export the item master (OITM) to a spreadsheet, have ChatGPT normalise names and draft purchasing descriptions, then review and import the changes through the Data Import/Export tool or Service Layer.
Explain a query result
Paste the output of a Query Manager report and ask for the three things that changed against last month, with no live connection at all.
Order status for sales staff
A custom GPT with one action that calls your own small API, which in turn reads order and delivery status from the Service Layer.
Draft customer emails
Give ChatGPT a delivery note and open order lines, and have it draft a clear delay notice that the salesperson edits and sends.
Write SQL and formatted search help
Ask ChatGPT to draft a Query Manager query against the B1 schema, then test it on a copy company before saving it.
Summarise supplier documents
Upload a vendor contract or certificate, ask for key dates and obligations, then record the result in a user-defined field after review.
SAP Business One integration surface
- Exports: Query Manager, Crystal Reports and the Data Import/Export utilities produce Excel or CSV that can be uploaded to ChatGPT with no integration.
- Service Layer: the REST/OData interface that a small wrapper API can call. Custom GPT actions need an HTTPS endpoint with an OpenAPI description, so the wrapper, not the Service Layer itself, should face the internet.
- SQLQueries and views via the Service Layer: let the wrapper expose only a few fixed, parameterised read shapes. Availability varies with version.
- Database read replica or reporting copy: a safer source than the production database when the wrapper needs heavier queries.
- Authentication: the Service Layer uses a login with company database, user and password and a session cookie, so the wrapper has to manage sessions and map GPT callers to a restricted B1 user.
How it works
- 1
Decide what level of access you need
Manual exports solve many cases. Only build a live connection when the same question is asked daily and the data is not highly sensitive.
- 2
Check your ChatGPT plan's data controls
Review whether the plan you use trains on inputs, what retention applies and who administers the workspace. Business and enterprise offerings generally differ from personal accounts, so read the current terms.
- 3
Build a thin wrapper API
Create a small service with a few endpoints such as order status or item availability. It logs in to the Service Layer with a restricted user and returns only a minimal field set.
- 4
Describe it with OpenAPI
Write an OpenAPI spec for those endpoints and register it as a custom GPT action. Keep descriptions precise so the model chooses the right call.
- 5
Secure the endpoint
Use an API key or OAuth, rate limits, IP restrictions where possible, and a gateway in a DMZ. Never open the Service Layer port directly.
- 6
Pilot with a small group
Give five users access for two weeks, review the logs for odd requests, and refine tool descriptions.
Example flows
Order status from a custom GPT
Trigger: A salesperson asks, "Where is order 4521?"
- The GPT calls the wrapper's order-status action with the document number.
- The wrapper queries the sales order and linked delivery and invoice through the Service Layer.
- It returns status, promised date and carrier reference only.
- The GPT replies in plain language.
Result: The salesperson gets an answer without opening B1 or calling the warehouse.
Weekly export review
Trigger: A finance user exports aged receivables to Excel.
- The file is uploaded to a ChatGPT workspace with data controls approved by IT.
- The user asks for accounts that moved into a worse ageing bucket.
- ChatGPT produces a summary and draft reminder wording.
- The user checks the figures against B1 before sending anything.
Result: A manual review that took an hour is done in minutes, with no system integration.
Keeping data on-prem
If your policy forbids sending B1 data to a third-party model, keep the same wrapper-and-tools design but point it at an open-weight model hosted on your own server. The wrapper and Service Layer stay inside your network, and users reach the assistant through an internal web page or chat client. You lose some model capability compared with the newest hosted models, but for status questions, summarisation and drafting a mid-sized open model is usually adequate. This is also the cleaner answer for customers with contractual limits on where pricing, drawings or part numbers may be processed.
Security checklist
- Do not expose the Service Layer to the internet; put a hardened wrapper in a DMZ.
- Return the minimum fields needed and strip cost, margin and personal data unless required.
- Use a restricted B1 user for the wrapper with read-only authorisations.
- Check your plan's training and retention settings before uploading any exports.
- Log actions and review them for unexpected queries.
Ways to get it done
Manual exports into ChatGPT
Effort: None, but depends on discipline.
Best for ad hoc summaries and drafting on non-sensitive data.
Custom GPT with a wrapper API
Effort: One to three weeks including security review.
Good for a handful of repeated lookups and a small user group.
Private model with a partner such as Netray
Effort: Scoped pilot, then expansion.
Appropriate when data may not leave your network or you need many tools and write-back with approval.
Pitfalls
- !Uploading customer price lists or drawings to a personal account with default data settings.
- !Exposing the Service Layer directly to make a GPT action work.
- !Trusting ChatGPT's SQL against the B1 schema without testing on a copy company.
- !Skipping user-defined fields so answers miss customer-specific data.
- !Letting a GPT action return full documents when a few fields would do.
Frequently asked questions
Can ChatGPT connect directly to SAP Business One?
Not out of the box. A custom GPT can call an HTTPS API you provide, so you would build a small wrapper in front of the Service Layer. Pointing ChatGPT straight at the Service Layer is not advisable because of authentication, session handling and exposure concerns.
Is it safe to paste B1 exports into ChatGPT?
It depends on the data and your plan. Review the provider's current terms on training and retention for your account type, and keep sensitive items such as cost, margin and personal data out unless your policy allows it. Many firms set rules for what may be uploaded.
Can ChatGPT create sales orders in B1?
It could if your wrapper exposes a write endpoint, but that is risky. A safer design creates a draft document that a user reviews and approves in B1, so approval procedures and stock effects remain under human control.
Why would I choose a private model instead?
Typical reasons are contractual restrictions, defense or aerospace data, pricing sensitivity, or simply wanting predictable costs. A private open-weight model keeps all queries inside your network, at the price of managing hardware and accepting some capability gap.
Do I need to be on HANA for this?
No. The wrapper talks to the Service Layer, which works across supported database editions. HANA mainly matters if you want heavy analytic queries directly against the database. Confirm Service Layer availability for your version and database.
Want ChatGPT working on your SAP Business One?
A free 30 minute assessment: we map the integration path for your version and setup, and where it should run.
Get a free assessmentTry it here, free
SAP B1 Service Layer Sales Order JSON Generator
Fill in server, company database, customer and line details to get ready-to-test Login and sales order POST requests for the Service Layer REST API.
New to SAP Business One? Start with the SAP Business One guide and toolkit.
Keep exploring
Free tools
- SAP B1 Service Layer Sales Order JSON GeneratorFill in server, company database, customer and line details to get ready-to-test Login and sales order POST requests for the Service Layer REST API.
- SAP B1 Implementation Cost CalculatorEnter users, rates and effort days to see first-year and three-year SAP Business One cost, using your own partner quotes.
- SAP Business One Service Layer Request BuilderFill in host, company and item details and get ready-to-paste Service Layer requests for login, item query, sales order and production order.
ERP fixes and how-tos
- Private AI for SAP Business One: Service Layer, a read replica and a local modelAdd a private AI assistant to SAP Business One: Service Layer reads, SQL Server or HANA replica, document index, local LLM, permissions and approvals.
- SAP Business One integration: Integration Framework (B1i), Service Layer, DI API and connectorsSAP Business One integration options compared: Integration Framework (B1i), Service Layer, DI API and SaaS connectors, and when to use each.
- SAP Business One production planning: MRP, production orders, cost and variancesSAP Business One production planning end to end: MRP, planned to released orders, cost rollup, variance reporting and when you need a manufacturing add-on.
Runs with your ERP
- SAP Field Service Management with SAP Business OneHow SAP Field Service Management, with Coresuite, syncs with SAP Business One service calls and billing: data flows, setup, pitfalls and a private AI layer.
- ECOSIRE CPQ Configurator with SAP Business One: guided quotingHow the ECOSIRE CPQ configurator uses option models, requires/excludes rules and live cost roll-up to create SAP Business One quotes and orders.
- Ecosire quality control and inspection with SAP Business One: CoA, NCR and CAPAHow Ecosire's quality add-on for SAP Business One handles inspections, certificates of analysis, NCR and CAPA ageing and supplier quality reporting.
AI on your ERP
- AI for SAP Business One, Without Sending Your Books to a Public APIAdd AI to SAP Business One without sending customer data or financials to a public API. Service Layer and DI API integration, sized for an SMB budget.
- AI for SAP Business One Purchasing, From Purchase Request to A/P InvoiceGround AI in SAP Business One purchasing: purchase requests, POs, goods receipts and A/P invoices via Service Layer, read-only, on-prem or private cloud.
Integrations
- How to connect ChatGPT to SAP S/4HANAConnect ChatGPT to SAP S/4HANA with custom GPT actions over OData, or with exports. Compare live access, enterprise data controls and a private model option.
- How to use ChatGPT with SAP Business ByDesignConnect ChatGPT to SAP Business ByDesign: export versus live access, custom GPT actions over OData, enterprise data controls, and when a private model is safer.