For service firms running Vertec

A custom Vertec MCP server for what Vertec's own MCP server does not cover

Vertec ships its own Vertec MCP Server, generally available since release 6.8.0.21 and pre-configured for Claude, ChatGPT and Copilot, as of 21 September 2026. Start there. A custom Vertec MCP server only pays off beyond it: aggregations through the BI API, questions across Vertec and another system, or a narrower tool set on the Vertec REST API.

Checked 21 September 2026

Vertec at a glance: interfaces, access and what exists for AI

What we found
InterfacesVertec MCP Server (MCP over the Vertec REST and ModelMetadata web services, docs public): from release 6.8.0.21; runs with every instance on Vertec Cloud Server, on-premise enabled via Vertec.ini · Vertec REST API (REST/JSON, docs public): from 6.8.0.17, delivered inactive and enabled under Settings > Extensions > Web Services; /objects/<classname> for read and write, and every query needs an OCL expression · BI API (REST/JSON, docs public): /api/bi with /getdata for measures by dimension as CSV or JSON, and /measures; the XML interface remains as a legacy option
AuthenticationREST calls carry a per-user API token (since 6.6) as a Bearer token; username and password authentication is disabled since 6.7. Data is filtered by the user's Vertec rights. The Vertec MCP Server uses OAuth 2 with PKCE and Dynamic Client Registration, with the Cloud Server as OAuth provider.
Vendor MCP serverVendor server, GA. The Vertec MCP Server is generally available since release 6.8.0.21 (30 July 2026) and on Vertec Cloud from 11 August 2026. It is generic over the whole Vertec model including customisations, can read and create objects, and is pre-configured for Claude, ChatGPT and Microsoft Copilot. Vertec says it is not suited to large object lists and recommends the BI module for measure analysis.
Vendor AI assistantVertec runs an AI support chat, pilots AI know-how management, offers receipt recognition and a Python LLM client, and ships the Vertec MCP Server.
Other connectorssimple-vertec-api is a community Node.js wrapper for the legacy XML web service; it is not an agent connector.

Vertec AG (CH). Project- and mandate-based service firms in the DACH region, such as law and tax firms, trustees, engineering and architecture offices, consultancies and IT service providers; Vertec states over 1,300 customers.

Why Vertec is hard for agents

Where a Vertec connection needs engineering, not configuration

01 · Overlap

Vertec already ships an MCP server

The Vertec MCP Server covers the whole Vertec model and is maintained by Vertec. A custom build only earns its place where that server stops: aggregations over large lists, which Vertec says its server is not suited to, questions that combine Vertec with another system, or a narrower tool set with fixed, reviewed operations.

02 · Query language

Every REST query needs OCL

The REST API requires an OCL expression on every query. Those expressions run against your own classes, custom fields and link types, so a tool that works in one Vertec installation can fail in another. Named tools fix the OCL per operation instead of letting the model write it.

03 · Versions

REST and MCP need recent releases

The REST and ModelMetadata web services need 6.8.0.17 or later and ship inactive; the MCP server needs 6.8.0.21. Older on-premise installations on 6.6 or 6.7 only have the XML interface and the BI API, which changes what can be built.

04 · Licensing

Measures depend on the BI module

Aggregated figures such as fees and minutes by project and period come from the BI API. What it returns depends on the BI module licence, so the aggregation tools are a licensing question before they are an engineering one.

What we build

A first Vertec tool list, written as the manifest your IT team signs

Built on what Vertec AG documents: projects, users and employees, activities, invoices, expenses, addresses and contacts, time entries (services), BI measures such as fees and minutes by project, employee and period. Backends named below are documented resources; "confirmed in blueprint" means the vendor documentation sits behind a login and the resource is checked against your release first.

# vertec-mcp-server: tool manifest, draft for blueprint week 1
# Every tool runs as the person asking. Nothing is written without a person confirming.
tools:
  find_project:         # code or name -> project with leader, status and client
    backend: REST /objects with OCL query, project class confirmed in blueprint
    mode: read
  get_project_services: # project, date range -> booked time entries and fees
    backend: REST /objects with OCL query, or BI /getdata; class confirmed in blueprint
    mode: read
  get_fee_measures:     # measures, dimensions (project, employee, month) -> aggregated figures
    backend: BI API /getdata
    mode: read
  list_open_invoices:   # client -> open invoices with amount and due date
    backend: REST /objects with OCL query, invoice class confirmed in blueprint
    mode: read
  draft_activity:       # contact, subject, date -> CRM activity proposal for a person to confirm
    backend: REST POST /objects, activity class confirmed in blueprint
    mode: propose
audit: every call logged with user, tool, arguments and the Vertec response

The manifest is the contract. A question that needs an operation not on it gets a new line and a review, not a wider permission.

Vertec holds the records. An assistant without a governed way in answers from whatever someone pasted into the chat. The connector is the governed way in: named operations, the person's own permissions, every call logged.

What the connector answers, and what it will not do

Questions it answers

  • How many hours were booked on project P this month, and how much is still unbilled?
  • Which open invoices does client X have?
  • Which employees have missing time entries for last week?

What it will not do

  • Read more than the asking user may read in Vertec; every call runs with that person's own API token, and Vertec filters the data by their rights.
  • Create or change a record by itself: the only write tool drafts a CRM activity, and a person confirms it in Vertec.
  • Duplicate the Vertec MCP Server: if a question is answered well by Vertec's own server, the custom connector does not get a tool for it.

When this is the wrong page

If you run Vertec 6.8.0.21 or later and your questions are about single projects, contacts or invoices, use the Vertec MCP Server: it is generally available, maintained by Vertec and pre-configured for Claude, ChatGPT and Copilot. This page is only for what that server does not cover, such as aggregations over large lists, questions across Vertec and another system, or a narrower tool set, and for older releases without it.

Start with the Readiness Blueprint.

Two weeks, €4,900 fixed price. Week one agrees the tool list with your Vertec owners and checks every backend against your release and licence. Week two tests whether a person's own Vertec permissions reach through the connector. You keep the manifest and a costed build plan, whoever builds it. All prices are net and exclude VAT.

What we need from you: a named Vertec administrator, a test or sandbox system we may call, and one business owner who can say which questions the assistant should answer first. The first production connector is usually a use-case pilot from €15,000; it runs in your environment, and the code is yours.

Sources: Vertec MCP Server · Vertec REST API · Vertec API token · Vertec BI API · Vertec forum: Vertec MCP · AI at Vertec · Vertec. Read 21 September 2026. Vendor documentation changes; verify against the current release.

Related

Questions Vertec teams ask before the first tool call

Does Vertec have its own MCP server?

Yes. As of 21 September 2026, the Vertec MCP Server is generally available from release 6.8.0.21 (30 July 2026) and on Vertec Cloud from 11 August 2026. It covers the whole Vertec model including customisations, can read and create objects, and is pre-configured for Claude, ChatGPT and Microsoft Copilot. Use it first; build only for what it does not cover.

Do we need extra Vertec licences or releases for an AI connector?

The REST API needs release 6.8.0.17 or later and has to be switched on under Settings > Extensions > Web Services; the Vertec MCP Server needs 6.8.0.21. Aggregated measures through the BI API depend on the BI module licence. Anything beyond that is a question to check with Vertec.

Can the assistant see data a user is not allowed to see in Vertec?

No. Each REST call carries the asking person's own Vertec API token, and Vertec filters the data by that person's rights. Two users asking the same question can get two different answers, and that is tested before go-live.

What does a Vertec connector cost?

The two-week Readiness Blueprint is fixed at €4,900: agreed tool list, backends checked against your release, identity tested end to end, costed build plan. The production connector is usually a use-case pilot from €15,000. Prices are net, excluding VAT.

Next step

Not a sales call. An architecture call.

Thirty minutes with the engineer who would build the connector, with your Vertec administrator in the room if you like.