Rechnungssammler is looking for suitable companies for its public beta
AI engineering for pragmatic SMBs
I build AI software for real operational bottlenecks.
I am Luca Knaack, an AI engineer and full-stack developer from Germany. I help small businesses turn repetitive work, messy inboxes and sales research into lean tools that people can actually use.
Talk directly to the builder
Value before automation
Germany-based hosting possible
Invoices, leads, inboxes, files, internal tools and integrations
Consulting, prototype and implementation without a handoff between sales and dev
Services
From vague AI idea to useful working software
I work best on small, sharply defined problems: one process, one user group and a visible improvement. That is where AI becomes useful faster than in broad strategy projects.
AI consulting for worthwhile use cases
We identify which processes are commercially worth automating and where AI can remove meaningful friction.
- Use-case prioritization for lean teams
- Feasibility and data-readiness checks
- Roadmaps focused on ROI instead of hype
Custom software with AI inside
I build systems that combine AI components, business logic, data flows and a usable interface in one coherent product.
- Full-stack implementation
- LLM-powered product features
- Self-hosted or cloud-based deployment options
Workflow automation
In many cases the biggest win comes from connecting existing tools with a clean layer of logic and AI, not from replacing everything.
- Inbox, file and CRM workflows
- n8n, Make or custom API integrations
- Gradual rollouts for smaller organizations
Project experience
AI only pays off when there is clear operational friction
Invoice automation
Capturing invoices from inboxes, understanding them, naming them correctly and storing them in a way that supports accounting handoff.
Lead qualification
Turning large company lists into qualified target selections with LLM-based filtering instead of manual review.
Pragmatic adoption
Smaller, focused improvements are often more valuable than a broad AI program with no operational anchor.
Products
Products as a workshop, not a showroom
I build my own AI products because they quickly reveal whether an idea survives real usage. Rechnungssammler is intentionally open as a public beta; CompanyScorer is a reusable qualification flow for concrete sales cases.
Rechnungssammler
Pulls invoices from inboxes, uses AI to understand them and stores them in a traceable structure. It is not positioned as a finished mass-market product yet, but as a beta for companies willing to shape the workflow with feedback.
CompanyScorer
Uses LLMs and web crawling to score large company lists against custom target profiles. Useful when a real longlist and a clear sales profile already exist.
Working model
Start small, build cleanly, use it for real.
I work best with companies that have a concrete bottleneck, short decision paths and a willingness to start with something operationally meaningful.
Clarify the problem
Where is time being lost today, and where does manual work still dominate?
Constrain the solution
We define data sources, risks, expected impact and the narrowest useful scope.
Build the system
From prototype to a working implementation that fits your daily workflow.
Improve with usage
Real feedback and real usage patterns drive the next iterations.
Testimonials
Client quotes and references
This section is ready for approved testimonials and will automatically appear once entries are added to the site configuration.
FAQ
Questions that actually matter before the first call
If you already have a specific process in mind, we can start there immediately.
Is this mainly consulting or implementation?
Both. I can help define the right AI use case and then build the actual software or automation behind it.
What if we do not have the perfect use case yet?
Then that is the first step. We reduce the process to a scope where value, data quality and risks can be judged realistically.
Can this connect to existing tools?
Yes. Common integration points include APIs, inboxes, file storage, CRMs, n8n and Make.
Is Rechnungssammler finished?
No. Rechnungssammler is running as a public beta. I am looking for companies with a real invoice workflow problem that are willing to give feedback.