mindthread.

AI Systems

We design, build, and operate AI systems that have to work inside real companies and products: reliable, observable, cost-aware, and ready for European production environments.

Glass prisms in Mindthread blue

Four ways we work on AI

Start where you are. Most clients begin with one and add the next once the first one runs in production.

AI Strategy & Adoption

For companies where everyone wants AI and nobody agrees where to start or who owns it. We rank the use-cases worth doing and set up the governance and roadmap to run them.

  • Use-case portfolio with a business case for each
  • Governance model with named owners
  • Adoption plan for the teams who will use it
  • Twelve-month roadmap
Who it is for
Executive teams and transformation leads
Example
The Production AI Adoption Flywheel
First step
AI readiness blueprint, €3,500

Knowledge Systems

Documentation, tickets, and runbooks turned into a knowledge layer that copilots and agents can answer from, inside the tools your people already use.

  • Knowledge layer over docs, tickets, and wikis
  • Retrieval tuned and evaluated on your data
  • Copilots in the tools your teams already work in
  • Access control, audit logging, and EU data residency
Who it is for
CIOs and heads of service
Example
ITSM AI Copilot: 60% less first-line workload
First step
Pilot on your real data

Agentic Process Automation

Operations agents that run documented procedures end to end, behind a governed API that checks every action before anything changes.

  • Operations agents for real procedures and workflows
  • A governed API that can refuse an action
  • Approvals inside existing tools such as ServiceNow
  • An audit log of every action and decision
Who it is for
Heads of operations and IT operations
Example
From Runbook to Governed API
First step
Use-case workshop

AI-Native Products

AI features built into your own product and codebase, with the evals and cost controls to ship them to your customers under your name.

  • AI features inside your existing codebase
  • Model orchestration with evals in the release pipeline
  • Cost and quality controls per customer
  • White-label delivery
Who it is for
CTOs at software vendors
Example
ITSM AI Copilot, shipped inside a vendor's product
First step
Pilot, then a fixed-price project

AI starts creating value when it survives real users, audits, cost ceilings, and model changes.

That is why the operating layer comes with every offer above, instead of being sold as an extra.

Built into every offer

Evals and observability

Quality is measured before launch and watched after it, on your data.

Guardrails and cost control

Every model call has a budget and a boundary it cannot cross.

Your cloud or tenant

GCP, AWS EU, or on-premise, with EU data residency and EU AI Act readiness.

Operations after launch

SLA-backed on-call, incident review, and monitoring as models and data change.

From a free demo call to a system we keep running

Each step is priced up front, and you only climb as far as you need. Most teams start with a free 30-minute call with an engineer who has shipped this before.

Assess

Work out what is worth building, and what it will cost.

Step 01 Most common paid first step

AI readiness blueprint

€3,500

fixed price, about 3 days

An audit of whether your data, systems, and processes are ready for AI, and where it would actually pay off.

  • Data and infrastructure review
  • Use-case scorecard
  • Six-month roadmap

Fixed price, remote or on-site.

Step 02

Use-case workshop

€4,500

fixed price, 2 days

A working session to pick the use-case worth building, sketch the architecture, and agree the scope. You leave with a fixed-price offer for the build.

  • Use-case shortlist and scoring
  • Architecture sketch
  • Fixed-price offer for the pilot

Offer within five working days.

Build

Prove it on your data, then run it in production.

Step 03

Pilot

€15,000–30,000

fixed price, 6 weeks

We build a working prototype on your real data, with Go/No-Go metrics agreed before we start.

  • Prototype on your real data
  • Go/No-Go metrics
  • Business case for your stakeholders

Cost credited against the project.

Step 04

Individual project

€30,000–75,000

fixed price

We ship one real workflow into your own cloud, behind a governed boundary that can say no. It runs in production from the first release.

  • One real workflow, live
  • Runs in your cloud or tenant
  • Governed boundary with audit log
  • Owned by your team afterwards

Pilot fee credited here.

Step 05

Foundation & Enterprise

Custom

scoped per engagement

When the first build proves out and you need the full system: more workflows, governance, and rollout across teams or sites.

  • Multiple use-cases and teams
  • Governance and MLOps platform
  • Phased rollout

Scoped and quoted with you.

Operate

Keep it running and improving after launch.

Step 06

Operate

from €3,000

per month, rolling

We keep it running and improving as models, data, and usage change, so quality and cost stay under control after launch.

  • SLA-backed operations
  • On-call cover and incident review
  • Quality, cost, and model monitoring

Optional. Month to month.

Most teams start with a free 30-minute call

You talk to an engineer who has shipped this before. No slides, no commitment.

Book a demo call