EinfachAI
AI automation for documents, portals and back-office workflows.
Overview
EinfachAI builds AI assistants, agents and automation workflows that take over clearly defined work between people, documents and software. It blends 15 years of software engineering and roughly ten years of complex e-commerce systems with agentic AI work that goes back to 2023, when the first agent frameworks appeared.
The approach is deliberately unglamorous. Instead of an all-purpose autonomous agent that is trusted with everything, EinfachAI designs narrow, auditable workflows: a bounded set of steps, an explicit rule for handing unclear cases back to a human, and a log of every action taken. Language models are used where they genuinely add value – reading unstructured documents, searching internal knowledge, or driving an interface that has no API – and conventional code, rules and workers handle everything else. The result is automation that a team can actually run in a business context, not a demo that breaks on the third exception.
Key features
- Document understanding — PDF, scan or email in, structured record out. Invoices, delivery notes and product data are read, validated against rules and passed to an ERP, CRM or line-of-business system, with implausible values routed back for human review.
- Internal knowledge assistants. An assistant searches approved internal sources, assembles the context for a case and answers with citations and access rules, so
- support, operations and review work starts from a prepared brief instead of a blank page.
- Browser and portal agents for systems without an API. Supplier, customer and admin portals that offer no usable interface are operated by a bounded browser or desktop agent. Actions are documented, deviations escalate to a person, and nothing runs blind.
- Cross-system coordination. APIs, business rules, classic queues and AI steps are combined into one flow, so work moves across several systems instead of being
- handed to a language model step by step.
- Digital Product Passport readiness. Architecture, GS1 Digital Link resolution and integration with Shopware, PIM and ERP systems for storing and serving immutable passport data – including the structured product data that autonomous AI buyers need.
- Data-protection-first deployment. Sandboxed environments, closed cloud tenancies or local hosting are chosen per project, and where it is reasonable, open source models run on European infrastructure instead of customer data leaving for a US API.
Pricing
EinfachAI starts with a free technical first check of the automation idea: a short assessment of whether the process is a sensible candidate, what the data and system landscape allows, and what a realistic pilot looks like. Project, pilot and maintenance work is then scoped and quoted individually, based on the systems involved and the volume of documents or cases to be handled. Workshop and code-audit formats are available as day based engagements. There is no self-service plan, because every engagement starts from the actual process rather than a fixed feature list.
Pros & cons
Pros
- Senior engineering depth behind every workflow – the automation is built and maintained by the developer who designed it, with no agency overhead between client and implementation.
- Exception handling is designed in from the start, so unclear cases go back to a person with context instead of failing silently.
- Unusually strong fit for e-commerce and product-data work: Shopware, PIM, ERP interfaces and product-passport architecture are core territory, not a side skill.
- Clear position on data: sandboxed development, European or local hosting where reasonable, and open-source models when they are good enough.
- German-language communication and delivery aimed at companies in Germany, Austria and Switzerland, working remotely.
Cons
- Not a plug-and-play product: there is no instant signup, trial account or self-service dashboard. Every engagement starts with a scoping conversation.
- Project-based pricing means no published price list – cost depends on the process, the systems and the data quality.
- Delivery is remote-first, so on-site workshop days have to be planned ahead rather than booked spontaneously.
- Deliberately narrow scope: EinfachAI will turn down automation requests where the process or the data does not support a reliable workflow, rather than build something that cannot be trusted.
Frequently asked questions
Anything repetitive where information already exists in documents, portals or internal systems: extracting invoice and product data from PDFs, preparing support or review cases, moving data between systems that do not talk to each other, or operating a portal that has no API. The test is not whether a process sounds like AI, but whether knowledge, data, tools and recurring decisions can be connected reliably.
No. The goal is to remove the routine portion of the work and keep people on judgement, exceptions and customer contact. Workflows are built to hand back to a human whenever a case does not match the defined rules.
Shopware and headless commerce setups, ERP, PIM and warehouse systems, PostgreSQL and MySQL, Node.js and TypeScript services, REST APIs and webhooks, and MCP-based tool integration. Where no interface exists, a browser or desktop agent is used instead.
Development is sandboxed and no business-critical customer data, credentials or sensitive code is sent to public cloud models. Deployment options include closed cloud environments and local or European hosting, including self-hosted open-source models.
The technical first check is free and takes a short conversation. After that, a pilot is scoped to a single process so it can be measured, then extended once the numbers are real.
Rate this tool
Click a star to rate EinfachAI.
