AI agents · ERP integration
Connecting AI to Priority and your ERP, on your own data
An AI agent can connect to Priority, SAP Business One or another ERP through the system's API. It starts read-only, and any write to the ERP waits for a person's approval. Feasibility for your version is checked in Discovery.
Discovery is a fixed fee, credited to your first months of monthly care
- For
- Finance, operations and IT managers at companies running Priority or another ERP
- The agent
- Answers questions from the ERP, drafts documents and reports, and asks approval for every write
- First step
- A Discovery, which also checks feasibility for your version and modules
What you can do
What can you do once AI is connected to Priority?
Examples of what an agent can do with your ERP. All of them start read-only, and every write goes through approval.
Plain-language questions on stock and orders
For example: how many units of a given item shipped this month. The agent answers from the ERP and shows where the figure came from.
A draft document from an email
A customer order or purchase request that arrives by email becomes a draft, which waits for approval before it's entered in the ERP.
Matching supplier invoices to orders
The agent matches an invoice to the purchase order and the delivery note, and sends only the mismatches for approval.
A daily exceptions report
Every morning: late orders, stock below minimum and open invoices. Only what needs attention.
Order-status answers for customers
The agent drafts a reply from the ERP and the production plan, and whoever owns the customer approves it.
Data clean-up suggestions
Duplicate customers, missing fields or items without master data. The agent suggests a fix, and a person approves it.
Connecting safely
How do you connect an AI agent to an ERP safely?
An AI agent connects to Priority through the system's API, using a dedicated user with minimal permissions. At first it only reads data; every write needs a person's approval and is logged.
- Read-only first
- At first the agent only reads: it answers questions and prepares drafts outside the ERP. The ERP stays the system of record, and none of its data changes.
- Writes only through approved actions
- When the agent starts writing, it does so only through actions defined in advance, and each one waits for a person's approval before it's entered.
- Every call logged
- Every call to the ERP, every draft and every approval is logged, so you can review afterwards what happened and why.
- Feasibility before commitment
- Feasibility for your version and modules is checked during Discovery, before any commitment to setup.
Built-in or custom
Priority's built-in AI or a custom agent?
The two don't compete, and they often complement each other.
ERP vendors, Priority among them, are adding built-in AI features to their systems. These work inside the ERP, on its data, and evolve on the vendor's roadmap. For common tasks inside the system, they're often the simplest route.
A custom agent also works outside the ERP: it connects the ERP to email, drawings, production data, accounting software or WhatsApp Business, and follows your own workflow and rules. You choose the model, cloud or local, and the approval points.
If the task lives entirely inside the ERP, check the built-in features first. If it runs across several systems and documents, a custom agent is the better fit. In many cases you use both, and you don't need to hire a full-time AI implementer to do it.
Priority and SAP are trademarks of their respective owners. The names are used here descriptively and don't imply any partnership or certification.
Systems
Which systems can an AI agent connect to?
Any system with an API, a database view or a clean export. These are examples, not a list of past projects, and feasibility is checked during Discovery.
ERP
Can connect to, for example:
- Priority
- SAP Business One
Office and messaging
Can connect to, for example:
- Microsoft 365
- Google Workspace
- WhatsApp Business
- CRM
Finance and data
Can connect to, for example:
- Accounting software
- SQL databases
- OPC UA, read-only
Method
From feasibility check to an agent working with your ERP
Five steps, each with a gate. API access, version and modules are checked in the first step, before any commitment to setup.
01
You start here
Discovery
We check API access, version and modules, map one process with real examples and estimate the return on investment.
No agent yet
Gate: You decide whether to go ahead with setup
02
Baseline
We measure how long each question, document or match takes today, how many there are each week and how many errors surface later.
No agent yet
Gate: You approve the baseline and the targets
03
Build
We build in a test environment, with a dedicated ERP user. Every write action gets an approval step, and every call the agent makes is logged.
Agent drafts in a test setup
Gate: The agent passes the test set on real data, read-only
04
Shadow mode
The agent drafts answers and documents alongside the team, but nothing is entered in the ERP. Drafts are compared with the team's work, for a period set in the Discovery report.
Runs alongside your team, sends nothing
Gate: Written criteria, for example: at least 70% of drafts approved without edits, and zero critical errors
05
Go-live
The agent answers and drafts documents, and every write to the ERP waits for approval. Autonomy widens only while most drafts are approved without edits.
Agent acts; you approve key steps
What you get at the end of each step
- Discovery
- Discovery report: process map, flow design, ROI estimate and a fixed quote
- Baseline
- Baseline figures the agent is measured against
- Build
- A working agent in a test setup, with an action log
- Shadow mode
- A side-by-side comparison of agent drafts and team work
- Go-live
- Monthly report: tasks done, time saved, errors caught at approval
Discovery
What do you get from a Discovery?
A fixed-fee, fixed-scope first step: one process, one report you can act on, and the fee is credited toward your first months of monthly care if you continue.
What you bring
- One process owner who knows the work and can go through examples with us
- 20–100 real past examples, such as emails, quotes or invoices, sanitised if needed
- Read access to, or exports from, the documents and systems the process uses
- Fee
- Fixed and agreed up front, in a written proposal
- Credit
- Credited toward your first months of care, if you go ahead with setup
- Data
- NDA on request; examples can be sanitised
- Monthly care
- Cancel any time with one month's notice, no minimum period
Discovery is a fixed fee, credited to your first months of monthly care
- 01Process mapHow the process runs today, step by step: who does what, in which systems
- 02BaselineTime per task, weekly volume, errors and response time to customers
- 03Agent flow designSteps, approval points, systems, model, and whether it runs in the cloud or locally
- 04ROI estimateHow much time the agent is expected to save, against the cost of setup and care
- 05Draft go-live criteriaWhat must hold in shadow mode before the agent starts acting
- 06Fixed quote for setup and careA fixed price for the workflow, set before any build work starts
Connecting AI to your ERP: FAQ
Can ChatGPT or Claude be connected to Priority?
Yes, through an agent that works with Priority's API using a dedicated user. The agent can use models such as GPT or Claude, and the model is chosen per task for the best price-performance. When the data is sensitive, a local model on your own servers is an option. Feasibility for your version is checked during Discovery.
Can the agent change data in Priority?
Only through actions defined in advance, and each one waits for a person's approval. At first the agent works read-only: it answers questions and prepares drafts outside the system. Writing is switched on only after shadow mode, and on criteria agreed in writing.
Does it work with SAP Business One?
An agent can connect to SAP Business One in the same way, through its Service Layer API: read-only first, with writes only after approval. Feasibility for your version and setup is checked during Discovery, before any commitment to setup.
Do we need Priority's AI module?
No. A custom agent works through the system's API. If you have the built-in features, the two can complement each other: the built-in features for tasks inside the ERP, and the agent for workflows that run across several systems and documents.
Where does ERP data go?
That depends on what you choose. We can use cloud models under terms that don't let the provider train on your data, or a local model on your own server, so the data never leaves your organisation. Either way, the model only receives the fields the task needs.
What if our version is old or heavily customised?
That's checked during Discovery. Where there's no suitable API, a database view or a clean export can sometimes be used instead, read-only. If there's no safe way to connect, the Discovery says so before any commitment.
How much does an AI–ERP integration cost?
The ERP connection is one of the cost drivers: how many systems need connecting, and whether they have a clean API. It's priced per workflow, as setup plus monthly care, and you pay the model usage directly. You get the fixed price in writing after the Discovery.
Pick one question you ask Priority every day, and we'll start there
Tell us which system and version you run, and what you'd like to ask it. An engineer will reply, and the Discovery fee is credited toward your first months of monthly care.
Usually a reply within 3 working hours, by email or phone