How we work

Every Obsidra project moves through five stages: Discover, Research, Design, Build and Support. Each stage takes a less certain input and hands a more certain one to the next.

Stages

Five stages

Not every project uses all five. A research study ends after Research; a build on a clear specification starts at Design. The order stays the same.

Every Obsidra project moves from a less certain input to a more certain output, one stage at a time.

Discover
What goes inA goal or a problem, described in your words
What comes outA written problem statement and a scope document.
What happens
We talk to the people who own the problem, read what already exists and agree the question.
What we need from you
Time with the decision owner and access to existing documents.
Research
What goes inThe agreed question
What comes outFindings, the evidence base and a recommendation.
What happens
Market, customer and data research, with every finding tied to its sources.
What we need from you
Introductions for interviews and access to internal data, where relevant.
Design
What goes inFindings and a recommendation
What comes outA specification the build is measured against.
What happens
User journeys, data model, architecture and, for AI work, the evaluation set.
What we need from you
Review sessions and decisions on trade-offs.
Build
What goes inThe specification
What comes outWorking software or AI features in production.
What happens
Short increments, each deployed to a staging environment and reviewed with you.
What we need from you
A product owner who can accept or reject each increment.
Support
What goes inA system in production
What comes outA stable system and the data for the next round.
What happens
Monitoring, fixes and planned improvements, with usage data feeding the next decisions.
What we need from you
Priorities for improvements and feedback from users.

Lengths sketch the usual share of effort for planning. Real projects vary, and many use only part of the line.

Responsibilities

Who does what

Each stage has work on our side and decisions on yours. Neither side can skip its lane.

Every stage needs something from both sides. The first thing we ask for is a person who owns the decision.

Engagement

Working models

The scope decides the model. All three are agreed in writing before work starts, without prices on this site because they depend on scope.

Fixed scope

When it fits: The question or the product is clear enough to write down in advance.
How it is priced: One agreed price for an agreed set of deliverables, split into milestones.
What you receive: The deliverables listed in the scope document, accepted milestone by milestone.

Best for research studies, readiness assessments and well-defined first releases.
Changes in scope are written up and agreed before they are built.

Dedicated team

When it fits: The work is ongoing and priorities will change as you learn.
How it is priced: A monthly fee for an agreed team composition, reviewed each month.
What you receive: Continuous delivery against a backlog you prioritise, with a monthly report.

Best for products after first release and for long data or automation programmes.
The team works in your tools and your repository from the first day.

Consulting

When it fits: You need an outside view, a review or a decision, not a build.
How it is priced: Time-based, against an agreed number of days or sessions.
What you receive: Written recommendations, reviews or architecture decisions, plus working sessions.

Best for technical due diligence, architecture reviews and AI strategy questions.
Recommendations are written so your own team can act on them.

After launch

The loop that follows

Once a system is in use, it produces evidence of its own. That evidence starts the next round.

Support is where the next round of research starts.

How to read it

Solid arrows: The order of work inside one round: decide, build, use, measure.
Brass arrow: Evidence from real use going back into the next decision.
Where it runs: The Support stage. Each turn of the loop is a small new project.

Prerequisites

What we need to start

Four things, before the first stage begins. The FAQ covers contracts, ownership and confidentiality.

A named person who owns the decision or the product on your side
The question or goal, even if it is still rough
Any existing material: research, data samples, documentation, access to systems
Constraints we must respect: deadlines, budgets, data location, internal policies

We sign an NDA before you share anything confidential, and the results of the work belong to you. Details in the FAQ

Next step

Ready for the first stage?

Tell us the decision or the system in front of you. On the intro call we say which service fits, what we would need to start, and whether we are the right team.

Book an intro call