Connected to whatyou already use.
An ERP does not replace everything. It has to connect to what stays: accounting, online shop, carrier, till signature, scale, government interface. fabular brings these integrations with it – and the open fabular.API for everything else.
fabular. process chain
Real timeOpen orders
128
Batches today
24
fabAI flagged batch L-2481: three days from best-before – FEFO suggestion created.
- Over 100 ready-made integrations
- Open REST API
- Cloud, on-premises, hybrid
How an integration comes about
Not every interface needs a project. The route depends on whether the integration already exists, whether it can be configured, or whether something really has to be developed.
Included ready-made
What is in the standard – DATEV, Shopify, DHL, DPD, FinanzOnline, A-Trust – is configured, not developed: enter the credentials, check the mappings, go live. Updates arrive with the release.
Configured through Chameleon
The middleware service connects external systems over REST, SFTP or FTP without any code being written. Endpoint, authentication and field mapping are maintained in the system, and the scheduler handles recurring runs.
Developed through the fabular.API
If the counterpart does not exist ready-made, the integration is built on the platform's REST services. Whatever turns out to be generally useful later moves into the standard.
Systems that fabular connects to
Over 100 integrations are part of the product, and new ones keep arriving from customer projects. The extract below shows the areas and some of the systems in each – which ones are active in your project depends on your landscape.
Finance and accounting
DATEV, BMD, RZL, SAP and Sage – journal entries, open items and master data move automatically instead of being exported and imported.
E-commerce
Shopify, Shopware, WooCommerce and marketplaces. Anyone using the included online shop needs no interface here at all.
Logistics and shipping
DHL, DPD, GLS, Österreichische Post, UPS, FedEx, DB Schenker and Dachser – shipment creation, labels, pre-advice and status feedback straight from the order.
Payment
PayPal, credit card, SEPA, TWINT and invoice, processed through SaferPay, CrefoPay, Qenta QPay or Wirecard – with authorisation, capture and status reconciliation.
Tax and authorities
EU VIES for VAT number checks, FinanzOnline in Austria, ATLAS for customs processing, Intrastat declarations and data exchange with Agrarmarkt Austria.
Till security
A-Trust RKSV for receipt signatures in Austria, Swissbit TSE for the German till security regulation.
Quality and certification
Laboratories, inspection bodies and QA systems such as AMA Marketing, BioC and AgroLab – test results end up on the batch, not in a mailbox.
Sign-in and SSO
Microsoft Entra ID via MSAL, Google and Facebook for shop logins, BankID, plus reCAPTCHA and device recognition at login.
Scales, tills and devices
Scales and till peripherals from Bizerba, Mettler Toledo and Avery through the PeriCom service, plus scanners, customer displays, cash drawers and silo sensors.
Production and plant
The shop floor data collection service takes in shift starts, batch information and events from production. Plant and machine controls via middleware, laboratory measuring devices via the WebSocket bridge.
EDI and file transfer
ORDERS, DESADV and INVOIC over SFTP or FTP, time-controlled through the scheduler. For retail chains with their own EDI network, our partner Procuros handles the pre-integration.
Master data
Article and partner data sources, classifications and catalogue formats – maintained once, valid everywhere.
Geodata
Google Maps Geocoding and OpenStreetMap for address resolution, map display and route planning.
Marketing and data
Klaviyo for email marketing, DataNature for data enrichment, signoSign for digital signatures.
Reporting
SAP Crystal Reports through the report server for document output, Apache Superset as the analysis interface with sign-in through fabular.
Infrastructure
Office 365 for mail and directory services, the Outlook plug-in for incoming correspondence, documentation systems such as BookStack.
Artificial intelligence
Azure OpenAI as the basis of fabAI – prompts, tool calls and analyses run through the platform, not past the system.
Everything else
Through automation platforms such as Zapier, Make or Power Automate, thousands of further applications are reachable without a dedicated integration. More on that below.
The platform's REST interface
The fabular.API is not an interface bolted on afterwards. These are the same REST services the interface itself works with – from sign-in through article and partner data to stock movement, order and invoicing.
- Base path /ACM/api/, JSON over HTTPS, JAX-RS in the Java backend
- Token-based sign-in with automatic renewal of the session
- Permissions per service through the same role model as the interface
- Sign-in via Entra ID, BankID or a dedicated user account
- fabular itself can act as an OIDC provider for neighbouring applications
- SOAP services are also available for older partner systems
Rights instead of full access
An API account gets roles like a user. Which service is released for which role is documented and traceable in the system.
Independent of deployment
The same interface in your own data centre, in the cloud and in the replication or offline operation of a branch.
For AI agents as well
Through the MCP server, AI assistants address the same services – with the same rights. More on that in the next section.
The detour that makes hundreds of systems reachable
When a tool is neither in the standard nor worth a project of its own, there is a third route: an automation platform as an intermediary. Zapier is the best known, and there are also Make, n8n and Microsoft Power Automate. These services are already connected to thousands of applications – from Google Sheets through Slack to CRM.
- An intermediary instead of individual integrations: you connect fabular to the platform once, not separately to every tool
- Rules instead of programming: the flows are clicked together in the browser – if this happens, then do that
- Both directions: fabular can trigger an automation and can itself be addressed by one
- One example: a new customer is created in fabular, which creates a record in the marketing tool and posts a message to the team chat
- A second: a form on your website is completed, which records the enquiry in fabular and assigns it to the responsible account manager
How it works technically
The platform calls the fabular.API when it needs to enter or read something. In the other direction, fabular sends a message through the Chameleon service to the address the platform provides – a so-called webhook.
What it suits
Peripheral processes that need to be up quickly: notifications, handovers to marketing or project tools, small data reconciliations. For bulk data and anything posting-relevant, a direct integration remains the better route.
What to keep in mind
Your data passes through a third party in the process. Check which fields really have to be transferred, where the service operates and what your data protection agreement says about it. We support you in weighing that up.
When the AI does not guess but looks it up
MCP stands for Model Context Protocol – an open standard for how an AI assistant accesses a business system. Without it, an assistant can only phrase what sounds plausible. With it, it looks up what is actually in the system, and may also enter something. fabular handles both sides of that standard: as an MCP server the platform is reachable for assistants, and as an MCP client it addresses the servers of other systems itself. An assistant therefore does not end at the platform boundary.
- The assistant reads real data: partners, orders, documents, figures – not an estimate, but the state in the system
- It searches documents by content, not by keyword: contracts, audit reports, manuals
- It may write where this is released – creating a CRM contact or leaving a note, for instance
- It can analyse and condense instead of only repeating figures
- It signs in like a user, via OAuth 2.1 with PKCE, and gets exactly that user's rights – no more
- Because every access goes through the same door, it is also traceable who queried what and when
- In the other direction, fabular uses the MCP servers of your other systems – an assistant can carry out a step there without a dedicated connector being built
- Which external servers are released is your decision; the release runs through the same permission administration as everything else
Not tied to one vendor
Because MCP is an open standard, the door works for different assistants – for the chat in fabular as much as for Microsoft Copilot or Claude. You decide what your people work with, without rebuilding the integration.
The door works both ways
As a server, fabular lets assistants in; as a client, fabAI goes out itself. A process that spans several systems needs both: an assistant that only knows its own house stops at the first system boundary.
The point where chains break
From order to paid invoice, four to six systems are usually involved. When the chain breaks, manual work appears: retyping, asking, reconciling. That is exactly where the API and MCP come in.
Staying auditable
An assistant that works past the system leaves no trace. Through the MCP server, every access stays tied to a user account, a permission and a point in time – the precondition for letting AI into audit-relevant processes at all.
Why openness decides in large processes
The larger the company, the more rarely a process runs in a single system. An ERP that closes itself off then becomes the bottleneck – and people fill the gaps between the systems with manual work. Two chains that have to run through.
Order to cash – from the order to payment received
- 1
Order comes in
Shop, portal, EDI
- 2
Availability and limit checked
ERP
- 3
Picked and labelled
Warehouse, scale
- 4
Shipment handed over
Carrier
- 5
Invoice transmitted
Invoicing, e-invoice
- 6
Payment matched
Bank, accounting
Six stations, at least four systems involved. Every transition without an interface means retyping, asking, reconciling.
Procure to pay – from the order to the payment
- 1
A requirement arises
Planning, recipe
- 2
Order goes out
Supplier portal
- 3
Goods receipt recorded
Warehouse, inspection
- 4
Invoice reconciled
Document recognition
- 5
Approval granted
Workflow
- 6
Payment instructed
Banking
The same logic on the purchasing side. Approval belongs in the workflow, not in an email distribution list.
What these two chains have in common: not a single step is difficult. The transitions are. An open interface makes sure that data flows at every transition instead of files – and that in the end it is traceable how a document came about.
This list is an extract. Whether your specific system is included – and at what depth – is something we clarify against your system list before we put an offer together.
On integrations and the API
Can fabular work with our existing accounting system?
In most cases yes. There are established integrations for DATEV, BMD, RZL, SAP and Sage, and everything else runs through the Chameleon service or the fabular.API. What is transferred – journal entries, open items, master data – and which system stays leading is defined in the project. Tell us the system and version and we will say specifically which route it will be.
How long does it take to set up an integration?
An integration that exists as standard – DATEV, DHL or Shopify, for instance – is usually ready for operation within a few days; the work is in credentials and mappings. Interfaces configured through Chameleon typically take a few days to two weeks. Individual development through the API depends above all on how well the counterpart is documented.
What if our system is not listed?
Then we assess the feasibility together. The precondition is an interface on the other side: an API, a file format over FTP or SFTP, or database access. There is a route in fabular for all three. Come to us with the specific system and you will get a reliable answer rather than a guess.
How is API access secured?
An account signs in and receives a token that is renewed automatically. Permissions run through the same role model as the interface – an account sees only what its role is released for. Microsoft Entra ID can be used for sign-in; conversely, fabular can act as an OIDC provider for neighbouring applications.
Do integrations also work in on-premises operation?
Yes. The platform runs in your own data centre just as it does as SaaS with us, and the interfaces are the same. The difference is in the network configuration, not in the function. For sites with a poor connection there is also replication and offline operation, for tills in a shop for example.
Do we get interface documentation?
Yes. For the agreed scope you receive the description of the services including role requirements. If your own developers build the integration, we set up access on the test system for that before it goes live.
Tell us about your system landscape
Say what should stay and what can be replaced. We check which integrations exist ready-made, what can be configured and where effort actually arises.