Automation around Priority ERP (an Israeli ERP) is built on its REST API, which is based on OData. When Priority is installed on-premises, the connection to the cloud goes through the on-premises data gateway, without opening inbound ports. On top of that connection you build Power Automate flows, Power Apps and Power Pages apps and portals that read from Priority in real time, and AI agents that read and write only through defined actions. The key rule: Priority stays the source of truth. Read live, store only keys, and sync only reference data.
What are the ways to automate around Priority?
- Power Automate flows: approval workflows, alerts, document generation and scheduled sync.
- Apps and portals: Power Apps for employees and Power Pages for suppliers and customers, with live data from Priority.
- Azure Functions: when you need complex logic, file processing or server-side token signing.
- AI agents: an agent that understands a request in natural language, pulls data from Priority and moves an action forward, through a controlled layer.
All of these rely on the same connection to Priority. Plan it once, as a shared layer, rather than building a separate connection for each process. You can read about the tools themselves on the Power Platform page.
How do you connect to Priority: REST API and OData
Priority has a REST API based on the OData protocol. It lets you query records with filtering and sorting, and create, update and delete records in forms, including subforms. A few points to know in advance:
- The Priority system administrator must open the forms the process uses to API access.
- Authentication is by password, Personal Access Token or OAuth2. Use a dedicated API user with minimal permissions.
- According to Priority's documentation, every record added or updated through the API counts as a transaction for licensing, and Priority Cloud also has a rate limit on calls per minute. That is another reason not to duplicate data without need.
In Power Platform, you wrap the API in a custom connector, which serves Power Automate, Power Apps and Copilot Studio.
What do you do when Priority is installed on-premises?
Many organizations run Priority on their own servers. This is where Microsoft's on-premises data gateway comes in: a component installed in the organization's network that opens only an outbound connection to the cloud, with no inbound ports. Through it, Power Automate and Power Apps read from and write to Priority. It has size limits on requests and responses, so you pull only the fields you need, in pages. When Priority runs in Priority's cloud, no gateway is needed, and the flows call the API directly through the custom connector.
In Ashtrom Group's quality control system, Priority sits behind an on-premises data gateway, and every call goes through a controlled server-side process. No secrets in the browser.
Read live or sync?
| When it fits | What is stored outside Priority | |
|---|---|---|
| Live read | When you need the current state at the moment of the question: customer, order, delivery note, complaint | Only the key, for example a complaint number |
| Scheduled sync | Reference data that changes slowly: suppliers, equipment, items | A copy of the required fields only, updated daily or on an event |
| Controlled write | When a process ends with an entry in the ERP: an approval, a document, a status update | A log of the action, who approved and when |
Duplicating ERP data creates two versions of the truth, and both have to be maintained. So the default is a live read, with sync only where it saves repeated calls for stable data.
Which processes are worth automating?
- Approval workflows: invoices, purchase orders and contracts, with a "waiting for my approval" queue and a direct action in Priority at the end of the cycle.
- Documents: generating a document from Priority data, an editing and signing cycle, and returning the signed document to the ERP.
- Operational events: a new delivery note is matched to the truck's location, a new complaint creates a quality event.
- Alerts: a validity date about to expire, a missing document, an approval waiting too long.
AI agent on Priority: how does it work safely?
An AI agent does not get free access to the ERP. It reaches Priority through defined processes or through an MCP layer that exposes only specific entities and actions, for example "get order by customer" or "create draft". Every write goes through validations and duplicate checks, and in sensitive cases also a user's approval. Every action is logged.
Start with reading, and add writing only where the process justifies it. One MCP layer can serve several agents, without building a separate integration for each.
Security and permissions
- No keys or passwords in the browser. Access to Priority goes through Power Automate or Azure Functions with Managed Identity.
- A dedicated API user, with permissions only for the forms and actions the process needs.
- Permissions on the app side too: each user sees only the data they are allowed to see, enforced on the server.
- A log for every read and write, so you know who did what and when.
For all connection patterns, including SAP, read the SAP & Priority integration page.
How do you start a Priority automation project?
- Pick one process: a process that repeats every day and runs today on Excel, email or the phone, such as an approval cycle or delivery tracking.
- Map the data: which Priority forms the process reads, what it writes, and who approves.
- Check the connection: Priority in the cloud or on-premises, an API user, forms open to the API, and a gateway if needed.
- Build reading first: a screen or an alert that shows real data, before the system writes anything to the ERP.
- Add controlled writing: with validations, approval and a log, and test in a test environment before production.
After the first process, the connection layer already exists, and every additional process is built faster.
Examples from our projects
AlgoSec · invoice approvals in Priority: a dynamic approval workflow in Power Apps, with direct actions in Priority.
Ashtrom Group · quality control: Power Pages and Dataverse with a live connection to Priority behind an on-premises data gateway. A quality event opens from a real chain of customer, site and complaint, only the complaint number is stored, and a daily sync covers about 421 mixers and 218 suppliers.
Ashtrom Group · digital contracts: 12 stages, 3 internal approval rounds and 33 automated processes. The contract document comes from Priority, and the signed contract goes back to the ERP.
Ashtrom Group · truck tracking: delivery notes from Priority are matched automatically to the GPS location by delivery note number, on one map with an arrival time based on traffic.
Ashtrom Group · AI agent on WhatsApp: the agent identifies the customer and the order, pulls the order, truck and driver from Priority, and returns an arrival time inside the conversation.
Frequently asked questions
Does Priority have an API you can build automation on?
Yes. Priority has an OData-based REST API that lets you query, create and update records in forms that the system administrator has opened to API access.
Can Power Automate connect to Priority installed on-premises?
Yes, through the on-premises data gateway, which opens only an outbound connection from the organization's network to the cloud, with no inbound ports. This is how Ashtrom Group's quality control system works.
Do we need to copy Priority data into Dataverse?
In most cases, no. Read what you need live and store only a key, such as a complaint number. Sync only reference data that changes slowly, such as suppliers and equipment.
Can an AI agent write to Priority?
Yes, in a controlled way. The agent reaches only the actions defined for it, through a process or an MCP layer, with validations, user approval before an entry and a log for every action. Start with reading.
Can we build an approval workflow that ends in Priority?
Yes. For AlgoSec, we built a dynamic invoice approval workflow in Power Apps, with direct actions in Priority.
Official sources: Priority Software: REST API documentation. Microsoft Learn: on-premises data gateway, custom connectors. Checked on 22.11.2026.
CTO at Zooz-Tech. Leads development and architecture of AI agents, Generative UI and Azure infrastructure.