Weekly, not quarterly.
Progress every week, not just at the steering committee. You see real interim results regularly and decide based on evidence.
Agentic Engineering
Custom AI for the mid-market
AI systems for work that
ties up expert time.
We build custom AI systems for high-volume, repetitive work — where off-the-shelf tools are too generic and classic automation is too rigid.
Free & no obligation · 30 minutes · One concrete process
We build and run our own AI products — and use what we recommend every day.
Where it pays off
Before the pilot comes the math.
We don't build because AI is possible. We build when a repetitive process has enough volume, effort, and data quality to make a custom system worth it.
Every pilot starts with a shared calculation: today's effort, a realistic share that can be automated, integration cost, operating model, and payback. If the math doesn't hold, we don't recommend a pilot.
Get done
Systems that take over steps of the work — not just draft answers.
Decide
Decision proposals where rules, context, and documents come together.
Connect
Connection to ERP, CRM, DMS, and line-of-business systems — where the work actually happens.
ROI calculation
Effort
Today · tied up in the process
With system · remaining
Investment
One-time · build
Ongoing · operations / month
Paybackin months
We run the numbers with you — based on your volume and your figures.
How we work
Close to the team.
And on the risk first.
In the loop from day one.
We work directly with the people who run the process today. That way we understand exceptions, tacit knowledge, and quality criteria early.
What could fail comes first.
Does the system read your documents correctly? Does it classify like your experts? Is the data ready to connect? If not, you know it in weeks — not in months.
How we deliver
Prototype, pilot,
production.
You decide after each phase whether the next step is worth it — from a prototype on a real case to a production system in daily operations.
PrototypeWeek 1 – 2
First understand,
then prove it.
We go into the process with your team, review real data, and build a quick prototype on a real case — before any budget is spent.
- Walkthrough with the team on real data
- Prototype as a feasibility test on a real case
- Business case: volume, effort, impact, payback
Kick-off & walkthrough
Data sample
Feasibility test
RecommendationBuild a pilot / no pilot
Today · manual
With pilot
Illustrative impact — the real figures come from the pilot, measured against your status quo.
PilotWeek 3 – 8
Build first
what could fail.
The smallest system that makes sense: real data, real workflow, in shadow mode alongside the live process — measured against your status quo.
- Narrow scope: one area, one use case
- Weekly iterations with real users
- Measured against the status quo: time, volume, errors
ProductionPilot + 2 – 3 months
Connect, scale,
hand over.
We connect the system to your existing systems, harden it for production volume, and hand it over cleanly to your IT. Operated by us, by you, or hybrid.
- Integration with ERP, CRM, DMS — SSO, audit logs, backup
- Training for the teams, handover with run-book
- Operations: managed, self-hosted, or hybrid — no lock-in
Custom system
ERP
Inventory, orders
CRM
Sales / service
DMS
Documents
SSO
Auth / directory
EU / on-prem · single-tenant · no lock-in
Unfuture AI Platform
We don't start from scratch.
The foundation is in place.
Custom systems run on our AI App Platform — single-tenant, self-hostable, and built so that your control over data and operations stays with you.
SSO and identity federation, connector framework, LLM gateway with budget control, full audit log. What every project needs, we don't rebuild every time.
Platform stack
Apps
Custom apps · database-per-app
Connectors
Mail · calendar · ERP integration
LLM Gateway
Multi-provider · budget control
Auth
SSO · identity federation
Hosting
EU / on-prem
Tenant
Single
Lock-in
None
What stays with you
Your data,
your code,
your control.
Code, data, and hosting — under your control
Single-tenant is the default. Source code, models, and data paths belong to you. Hosting in a European cloud or on-prem with you — your call.
Compliance in the architecture
EU AI Act, GDPR, and your industry-specific requirements sit in the system — not in an appendix. Whatever needs to be checked is checked traceably.
People in the lead where it counts
Human-in-the-loop for sensitive steps. The system proposes, your team approves — and every decision is traceable back to the source.
What we build when a
standard tool isn't enough.
A custom system for a specific client process — and two products of our own that grew out of exactly this way of working.
Use Case
Jansen: understanding tender specs faster
An internal tool flags the relevant door items in the bill of quantities, so sales can decide faster and review more cleanly.
Spin-off
ProductLens: customer feedback becomes product strategy
Pia brings together reviews, support, social, surveys and market data and translates customer voices into decisions for product, marketing and competition.
Spin-off
HeyVera
HeyVeraHeyVera: taking phone stress out of property management
Vera answers calls, asks follow-up questions, documents cleanly and turns it all into cases the team can work on.
Let's talk
30 minutes.
One concrete process.
Tell us about a process that ties up time, creates errors, or limits growth. We'll assess whether a custom AI system makes sense — and what a solid pilot would need to look like.