What is an MCP server for weclapp?
Typing up quotes, looking up order status, checking item prices, correcting master data, clicking reports together — in weclapp the bottleneck is rarely missing data but the manual work around it. This is exactly where an MCP server for weclapp comes in: it gives AI models such as Claude, GPT or Gemini a controlled line into your weclapp instance via the Model Context Protocol, so that the agent accesses the weclapp REST API in a standardised way — create quotes, check orders, read items and contacts, evaluate project hours — without you building your own scripts or fragile Zapier chains for every AI connection. What distinguishes MCP from a classic API integration is explained in our guide MCP vs. API & function calling.
weclapp is built as an all-in-one cloud ERP for German mid-sized companies: inventory management, CRM, accounting and project management share one data model. For an AI agent this means it does not have to reconcile several systems — an order, a customer, an item exist exactly once. This consolidation is the lever that makes automation particularly worthwhile here.
Typical visitors to this page run a growing company on weclapp and notice that a lot of time goes into routine work: typing quotes, checking status, correcting master data, clicking reports together. An MCP server brings this work to the agent — with clear control over what it may read and write.
In one sentence: weclapp + MCP means: you dictate a meeting note, the AI agent turns it into a quote draft in weclapp, checks item prices and customer data — and only creates it as binding after your approval. Designed in line with data protection, EU-hosted, with a log.
Important distinction: Community MCP servers for weclapp already exist (such as codelev/weclapp-mcp and generic OpenAPI adapters like AnythingMCP). They are good for trying things out — but not a production setup. A live system needs a tightly authorised technical user, EU hosting, error handling, logging and an approval workflow for write actions. That is exactly the gap we close.
Use cases: what can a weclapp MCP server do in concrete terms?
- Quote creation from meeting notes After a customer meeting, you speak or type a short note. The agent recognises customer, line items and quantities, matches items and prices against weclapp and creates a quote draft — you review once and send.
- Order status queries in natural language "Which orders from customer Müller have not yet been delivered?" or "What is due for invoicing this week?" — the agent pulls the answer directly from weclapp, without you filtering through lists.
- Automatic master data update New contact details from email or the inbox are matched against weclapp. The agent reports a missing VAT ID, IBAN or deviating addresses and proposes a correction — master data maintenance without double entry.
- Project reporting & analysis "How many hours are booked on project X, and what is the margin?" The agent condenses project times, orders and item costs into a short analysis — ideal for the weekly status view.
- Item & stock queries Before making a commitment, the agent checks availability, price tiers and delivery times of an item in weclapp and returns a reliable statement — instead of gut feeling in the sales conversation.
- Chase-up & follow-up routines Open quotes with no response are detected after a deadline; the agent prioritises by value and probability and writes follow-up emails as drafts — staggered, friendly, prepared for approval.
How the integration works
Architecture for weclapp:
AI agent ↔ MCP server (your EU host) ↔ weclapp REST API (token authentication, JSON)
Concrete technical steps:
- API token & role: According to weclapp, role permissions also apply to the API (weclapp documentation). We therefore create a dedicated technical user whose role releases only the areas your use cases need (e.g. quotes and items, but not the complete accounting), and use its API token. In addition, our MCP tools offer only the entities the use case needs. The token is stored encrypted on the EU host.
- Tool mapping: We define MCP tools such as
list_orders,create_quotation_draft,get_article,update_contact,get_project_hours. Each with a JSON schema, validated inputs and clear error messages. - Pipeline & data flow: Inputs (meeting note, email, question) are processed in a structured way and mapped to the right weclapp endpoints — including plausibility checks against existing master data.
- Separating read and write: Read queries run directly. Write actions (quote, order, master data) are created by the agent as drafts by default; committing them requires human approval — configurable per tool.
- Logging: Every API action is logged with timestamp, triggering user and diff — traceable who initiated what and when.
If you are planning the overall concept for your system landscape, you will find the service view on our money page MCP integration agency.
GDPR, hosting & security
weclapp is a cloud ERP; we check the hosting location and DPA against your weclapp contract. To keep the whole chain of ERP, MCP server and AI model compliant with data protection, we pay attention to the following for every weclapp MCP server:
- EU hosting of the MCP server (e.g. Hetzner or the Azure region Germany West Central in Frankfurt), no US edge in our part of the chain.
- weclapp API token stored encrypted, bound to a technical user with a minimal role and a defined rotation and revoke strategy — a compromised token opens only what this role may do.
- DPA with the AI provider with EU data residency (Claude via AWS Bedrock in the EU Frankfurt region, OpenAI Enterprise EU, Azure OpenAI in an EU region).
- Data minimisation in an all-in-one ERP: Because weclapp holds inventory management, CRM and accounting in one data model, restriction is particularly important here — through the technical user's role and the tools offered, the agent sees only the approved areas (such as quotes and items), not personnel costs, bank data or the complete financial accounting.
- Client and location separation in weclapp: If several companies or locations run through one weclapp environment, we tie the agent's role and tools to the respective client structure, so that no location sees another's data.
You can read more about the legal framework and protective measures in the guide MCP server: security & GDPR.
What does it cost?
Because weclapp brings a modern REST API and the data model is central, a first use case can be implemented quickly. Concretely, that means:
- First use cases productive quickly — typically quote creation or order status queries as the start.
- Clear scope per use case, no hours roulette. Scaling after that in small, clearly defined steps.
- Clearly delimited scope: You pay for connecting the weclapp entities your use cases need — not for a complete project across all modules.
We clarify the concrete figure in the initial consultation — depending on the number of use cases, the weclapp entities needed and the approval workflow. If you first want to clarify whether a custom MCP server is the right route at all, the guide Having an MCP server developed helps.
Related guides
Frequently asked questions
Is there a ready-made weclapp MCP server yet?
Yes, community projects exist — such as codelev/weclapp-mcp and generic OpenAPI adapters like AnythingMCP. They are suitable for trying things out. For production operation you additionally need a tightly authorised technical user, EU hosting, robust error handling, logging and an approval workflow for write actions. That is exactly what we set up.
Which weclapp data can an AI agent read and write via MCP?
Quotes, orders, items, contacts and projects are reachable via the weclapp REST API. For reading we use this for reporting and status queries, for writing for quote drafts or master data updates — critical actions go through an approval step.
How secure is the API token access?
The weclapp API token is stored encrypted on the EU host. What it may do is determined by the user's role, because according to weclapp permissions also apply to the API; we therefore use a dedicated technical user with a minimal role. Write actions run as drafts by default and are approved before they are committed.
Can weclapp + MCP be operated in line with data protection?
Yes, if the chain is right: hosting location and DPA with weclapp checked, the MCP server on an EU host, the AI provider with EU data residency and a DPA. Because role permissions also apply to the API according to weclapp, the agent works through a dedicated technical user whose role opens only the approved areas.
Which AI models can be connected to weclapp?
In principle any MCP-capable model — Claude, GPT, Gemini or Microsoft Copilot. The beauty of MCP: we define the weclapp tools (quotes, orders, items) exactly once, after which each of these models addresses the same interface. If you later change provider or combine several, the weclapp connection stays unchanged — you swap one model out and another in, without rebuilding the integration. The guide MCP vs. API & function calling goes into the mechanism behind this in more depth.
Do you want an MCP server for weclapp?
A free 30-minute initial call — we check which weclapp use cases are worthwhile and estimate the project effort.
Book an AI Check arrow_forward