Custom AI
For product companies that want AI as part of their product. We build LLM applications, RAG systems over your data, agentic workflows and computer vision models. The client gets a working system + source code.
AI into the company
For companies with 10 or more employees, where routine eats up hours every day. We do a process audit, find the top 3 use-cases with the highest ROI and deploy them, either off-the-shelf tools (ChatGPT Enterprise, Claude for Work, Glean) or custom AI assistants for HR, finance, support, sales. For smaller companies we have a dedicated bundle: AI and automation for small and medium businesses.
We try it on ourselves first
Every solution we offer to companies runs in our own company first. At omnius today we run:
- An AI phone system that routes calls and, with the caller's consent, transcribes them, so no phone request gets lost.
- Email requests as tickets in omniHub, each with an assigned owner.
- Slack digests and reminders daily, weekly and monthly, including a flag for any ticket nobody picked up in time.
- Meeting transcripts and receipt processing directly in the CRM.
It saves four people in management an hour a day, roughly €41,000 a year at market salaries. The details are in our case study. Next we are preparing AI that will talk to callers, and ticket triage with automatic priority levels.
We don't just hand over a tool. We take care of it for you.
Typical assignments we handle
- AI phone system and request handling, call transcription and email requests automatically logged as tickets with an assigned owner.
- Internal RAG assistants, a chatbot over internal documents, guidelines, historical reports. Employees get an answer in seconds instead of searching through 200-page PDFs.
- Document data extraction, automatic extraction of data from invoices, delivery notes, contracts, forms. Document → structured data in ERP or CRM without manual rewriting.
- Ticket classification and routing, AI analyzes incoming emails/tickets and routes them to the right person or suggests a reply.
- Predictive maintenance, a model over telemetry (typically via KALM:IT) that alerts the service team hours before an actual equipment failure.
- AI agents in the workflow, an agent integrated into omniHub or Procurly that reaches out to suppliers on its own, drafts price quotes or prepares documentation.
- Custom Copilots, domain-specific assistants for specific roles (branch manager, project manager, controller).
The room is large: email takes up about 28% of employees' working time and searching for information nearly another 2 hours a day (McKinsey), and IDC reports up to 2.5 hours a day spent searching. McKinsey estimates that today's AI can automate work activities that absorb 60 to 70% of employees' time. If automation takes over 2 hours a day for a manager with a gross salary of €3,000, that is roughly €11,200 a year per person.
Concrete results depend on the starting point, typically we see 15–35% time savings on automated processes and dozens of hours per month freed up for work that can't be automated.
How we approach security
- Local models (Llama, Mistral, self-hosted) are the default wherever data must not leave the perimeter, typically healthcare, public sector, energy, financial sector
- Cloud models (Claude, GPT) where speed and content quality matter and data sensitivity allows it
- SovereignAI where you need the power of large language models but frontier providers are effectively off-limits for public administration and regulated companies (data leaves the EU, GDPR guarantees are missing, the EU AI Act adds obligations). We deploy powerful models and inference capacity operated inside the EU, compliant with GDPR and the EU AI Act.
- Always: audit trail of every AI operation, GDPR-compliant logging, content marker for AI-generated output, hallucination control via retrieval grounding
We are not trainers
We can prepare an AI training, but a training on its own rarely changes anything in a company. We guide the company through the whole journey: from a PoC or pilot, through deployment into operation, to long-term care and further development. We want to be a long-term partner, not a one-off delivery.
Want to know where AI brings the most value in your company?
Send us a list of your routine tasks, a few bullet points or a single e-mail is enough. We'll come back with a proposal of what is worth automating first and where AI delivers the most value in step one.
