Office with 50 mails per day
AI categorises incoming mails (inquiry, order, support, spam), prioritises by urgency and writes draft replies for standard cases. Employee reviews, clicks send. From 2h of email care per day to under 30 minutes.
AI Integration
AI where it actually helps. Not as a toy chatbot.
We extend existing systems with AI: email inbox with AI drafts, WhatsApp bot with company context, document extraction, knowledge base with natural search. Measurable ROI, no gimmicks.
What you get
AI modules your team uses from day one
We don't build another ChatGPT clone on your website. We build modules that save concrete time.
- Email inbox with AI draft
- Incoming mail is categorised and prioritised, standard replies come as drafts to approve. Outlook, Gmail or Fastmail.
- WhatsApp bot with context
- Not the classic keyword responder. The bot knows your website, products and FAQ. Escalates to you on complexity.
- Document extraction
- Invoices, contracts, delivery notes, vehicle registrations. AI reads PDFs and photos and stores data structured. Manual entry disappears.
- Knowledge base with natural search
- Instead of Slack ping-pong: internal questions are answered directly from documents, wiki and meeting notes. With source citation.
- CRM with AI summary
- Meetings, calls, emails are automatically entered as notes in the CRM. You have the summary after the call, not after two days.
- Monitoring and A/B testing
- We measure accuracy, time saved, user satisfaction. If a component doesn't help, out with it. No AI for AI's sake.
Who this is for
Concrete SMB setups
Trade business with WhatsApp chaos
Bot handles 80 percent of standard questions 24/7 (hours, price range, appointment booking), escalates only on complexity or complaints. Owner has quiet evenings, customer gets an answer in 10 seconds.
Service provider with PDF flood
Incoming invoices, contracts, quotes are read automatically. Amounts, deadlines, parties land structured in the CRM. 40 minutes of manual entry become 5 minutes of review.
Agency with Slack overload
Knowledge base from Notion, Google Docs and past Slack threads. New employees ask the AI instead of colleagues. Source citation included, no guessing hallucinations.
Online shop with complex catalog
Natural language product search: 'Sofa under 800 euros in blue, deliverable before Easter' finds the matching items. Average conversion lift 8-15 percent.
How we work
Measure first, then build
- 01
Assessment
Where does your team lose most time? We examine processes and prioritise by expected ROI, not by hype factor.
- 02
AI selection
OpenAI, Anthropic or local models depending on data protection and budget. We're not tied to one vendor.
- 03
Integration
2 to 6 weeks depending on scope. We test with real data before production rollout. Weekly intermediate states.
- 04
Training and rollout
Onboarding for your team, feedback loop, iterative improvement. We stay close for the first 4 weeks.
Pricing
From EUR 2,500 setup plus EUR 99 to 299 per month
Entry with one module from EUR 2,500 one-off, monthly running costs depending on traffic and AI usage. Larger integrations or multiple modules on request.
See pricing and packagesFAQ
Frequently asked
Which AI model do you use?
Depends on the case. Claude for complex tasks, GPT-4 for volume tasks, local models (Llama, Mistral) for data-sensitive applications, Aleph Alpha when EU exclusivity is mandatory. We're not tied to one vendor.
Is this GDPR-ready?
Yes. For cloud AI (Anthropic, OpenAI) there are GDPR contracts and EU hosting options. For highly sensitive data we recommend local models on your server or our EU infrastructure.
What if the AI answers wrongly?
It happens. That's why we set guardrails: source citation, confidence scores, fallback to humans under uncertainty. Plus continuous monitoring: we see when something drifts.
Who trains the AI?
We usually don't retrain but use RAG (Retrieval Augmented Generation). Your documents, wiki, meeting notes are fed as knowledge base to the AI model. Advantage: updates need no retraining.
Does this work with my existing tools?
Most business tools (Microsoft 365, Google Workspace, Slack, Notion, Salesforce, HubSpot, Pipedrive, Shopify) have APIs or MCP connections. For specialist software we look individually.
Can I measure ROI?
Yes. We define metrics at project start (time saved per employee, response time, accuracy) and measure monthly. If a module doesn't deliver, we rebuild or drop it.
Next step
30 minutes of honest check.
We look at where AI really helps in your operation and where it's just buzzword. No sales pressure, only honesty.