hazelheartwood
EN
  • English
  • Nederlandssoon
  • Françaissoon
  • Españolsoon
  • Românăsoon

Home / How we work

How we work

Five principles we never compromise on.

You have probably seen them: AI projects that look impressive and create almost nothing real. The difference is transformation discipline: we find the one place where AI really changes something and we build only that, all the way to the users. These five principles decide how we compose a team, where we start, how we grow a tool, who owns it and when we call it done. Each one is a promise you can hold us to.

LeanteamsGemba-firstStaircaseapproachEmbedded inyour teamMeasurableimpactBUILT ONLEAN PRINCIPLES
  1. Lean teams, big impact
  2. Gemba-first, technology second
  3. Staircase approach, not a big-bang transformation
  4. An extension of your team, not another dependency
  5. Measurable impact is the only metric
Principle 1 of 5

Lean teams, big impact

One small team, embedded in your organisation and composed for the impact needed: your people, ours and our AI agents in the same circle. How we compose it is our concern; the impact is yours. Less waste means value for money, high focus and fast results.

people: yours and oursAI agents
One lean team, composed for the impact needed: your people, ours and our AI agents, in the same circle.

Seen in

Impact · Inside HazelHeartwood

A lean coach with three decades of know-how, for every project member

Principle 2 of 5

Gemba-first, technology second

Your problem lives where the real work happens. That is where we start, understanding the real pain point before any tool is considered. If we cannot define the value, we do not start the build.

Learning as a way of working

Understand, prioritise, define. Then build, test, learn, refine. Then scale.

The learning happens at the gemba, with the people who do the work. Only once we understand the problem and the scope is clear do we start the build.

UNDERSTANDPRIORITISEDEFINE the problemby impactthe scope MVA MINIMUM VIABLEACCURACY THE INTERCHANGE Deliver value early.Validate in the real world. BUILDTESTLEARNREFINE EXPANDSCALE scopeon proven pull SCOPE LINEunderstand before buildITERATION LINEMVA and continuous improvementSCALE LINEonly proven value earns the right to scale

Where we start instead

  • Not with a big-bang transformation or a technology-led roadmap, but with the one place where AI really changes something.
  • Not with a heavy enterprise-wide data platform before value is proven, but with the data the process already produces.
  • Not with a lab-only AI experiment, but with real users in real operations, from the first mock-up.

Seen in

Impact · Automotive

An AI agent for product strategy and planning, supporting 400+ planners

Impact · Retail

An AI agent that finds alternative suppliers for own-brand products, across Europe

Impact · Retail

Search results that finally match the query

Impact · Industrial manufacturing

Root cause in seconds, not hours, for every technician

Impact · Industrial manufacturing

Every order and every stock level, visible without touching the core systems

Impact · Social

City Pirates: less administration, more time with the members

Impact · Industrial manufacturing

From scattered costs to managed transparency

Impact · Social

Dialogz: a coaching tool built with employees with autism

Principle 3 of 5

Staircase approach, not a big-bang transformation

Tangible outcomes at every step, not a promise at the end. We combine rapid experimentation with regular feedback, and we only scale what has proven business pull.

The staircase

Value rises with how much the process changes.

Every step is available today, each one is worth more than the last and the steps are not delivery milestones. They are levels of how much the work itself changes, each resting on a deeper technical foundation than the last.

VALUE ↑ · HOW MUCH THE PROCESS CHANGES →Assistgeneric, everyday workmodel access, safety rulesSolvecustom-built, one step+ one clean dataset+ one system connectionAutomateagents act in the process+ live, trusted data+ write access to systems+ monitoring and loggingRedesignrebuilt around agents+ same data across markets+ shared business context+ coordination of agents, oversightTECHNICAL FOUNDATION NEEDED: DATA AND SOLUTIONS

The fourth step is gated by foundation, not by ambition or budget: the same data across markets, a shared layer of business context, coordination between agents and clear rules. Nobody can buy their way onto it, which is why it is defensible. The foundation built today is an advantage competitors cannot copy overnight.

Seen in

Impact · Automotive

One data foundation for vehicles, products, dealers and accessories across Europe

Impact · Automotive

From bespoke extractions to a governed data mesh

Impact · Industrial manufacturing

Every order and every stock level, visible without touching the core systems

Impact · Professional services

A data architecture that grows with every acquisition

Impact · Social

Tribe & Thrive: wellbeing that lasts longer than a workshop

Impact · Social

Dialogz: a coaching tool built with employees with autism

Principle 4 of 5

An extension of your team, not another dependency

Transformation is a capability, and it should be yours. Your people are co-owners from day one; we transfer capability through co-delivery and shadowing, so you keep transforming after we leave.

people: yours and oursAI agents
A flexible, cell-based structure of lean teams inside your organisation, that adapts as the work changes. The capability stays when we leave.

What every engagement produces

  • A value case and a KPI you already steer by, agreed before the build.
  • A tool in real operations, at an accuracy the business agreed in advance.
  • A team on your side that owns the backlog, the tool and the next step.
  • A measured impact and a time to value, told in the same six-step story as our other impact stories.

Seen in

Impact · Automotive

An AI agent for product strategy and planning, supporting 400+ planners

Impact · Inside HazelHeartwood

A lean coach with three decades of know-how, for every project member

Impact · Retail

One list of every AI tool in the group

Impact · Social

Experience@Work: late careers that keep giving

Principle 5 of 5

Measurable impact is the only metric

A tool nobody uses is a waste of your money. Every initiative is linked to a value KPI, end users are involved early, behaviour change is measured and we sign off only when the new way of working is your standard way of working.

Minimum viable accuracy

An AI tool that is almost right is not right.

ADOPTION ↑ · ACCURACY → minimum viable accuracy “right most of the time”: users check every answer, the solution adds a step users stop double-checking adoption starts, impact starts the bar is set by the business, before the build, per use case
A tool that is right most of the time sounds like a success and behaves like a risk: users cannot tell which answers are wrong, so they check all of them. Above the bar, they stop checking, and that is where adoption and impact start.

Minimum Viable Accuracy, or MVA, is the accuracy level a tool must reach before it goes anywhere near real operations, agreed up front with the business for each use case. Ideas become proofs of concept, and successful proofs of concept become minimum viable products with a clearly defined MVA. Only tools that have reached their MVA are operationalised and scaled, and only tools with proven demand and business pull get that far.

Acceptance criteria: data-retrieval accuracy, business accuracy, consistency and intuitiveness. Where the bar sits is for the business to decide, not the data team.

Seen in

Impact · Automotive

One data foundation for vehicles, products, dealers and accessories across Europe

Impact · Retail

An AI agent that finds alternative suppliers for own-brand products, across Europe

Impact · Retail

Search results that finally match the query

Let’s talk

Bring one of your own use cases.

We walk you through the routine. A first conversation costs nothing but an hour.

Let’s talkAll impact stories