How we work.

The working cycle

How improvement happens.

Five steps, repeated. Larger systems tend to emerge from successive passes rather than from one big design up front.

Understand

We sit with leadership and with the people doing the work. Where does a job wait on someone? What gets re-entered? What does nobody have visibility into until it is late?

Prioritise

We pick one opportunity and agree on the measure that defines success, where it stands today, and what would count as a real improvement.

Build

We design and build the system, then connect it to the workflow people already use. Adoption is part of the build, not a phase afterwards.

Measure

After a period long enough to mean something, we review whether it is being used and what moved against the baseline — including when the answer is less than we hoped.

Continue

What we learned changes the ranking. The next improvement is chosen from a roadmap that now knows more than it did.

Prioritisation weighs four things:Potential impactFeasibilityData readinessOperational risk

Capabilities

What we build, described the way the business would describe it.

The stack matters, but it belongs in the case studies. These are the four kinds of work an engagement tends to involve.

AI strategy and operational consulting

Opportunity discovery, workflow assessment and implementation planning. We map how work actually moves through the business, rank where AI and automation would earn their place, and write down what each one depends on.

Delivered as a plan you can act on with or without us.

AI automation

Systems that take on repetitive work: reading and answering messages, drafting and generating documents, moving a job from one team to the next without anyone re-typing it.

Handoff rules are set by you. Anything consequential stops for a person.

Custom systems and integrations

Software that connects the tools you already run, and purpose-built systems for the workflows nothing off the shelf fits. Built in your accounts, documented, and transferable.

We look for something existing first. Custom is the answer when it genuinely is.

Data and decision support

Reporting and analytics that answer the questions leadership is actually asking — built on the same data operations runs on, so the numbers reconcile.

Fewer dashboards, better ones.

Ways to work together

Most of our work is a standing relationship. Some of it isn’t.

Ongoing improvement partnership

Our preferred engagement

We keep a prioritised roadmap with you and work through it in scoped improvements, reviewing each one before choosing the next. Over time the individual pieces tend to become a larger system, but we don’t start by designing that system — we get there by building the parts that were worth building.

Support and development responsibilities are defined in the agreement, so it is clear what is covered and what would be new scope.

  • A live roadmap, re-ranked as we learn
  • Scoped improvements with an agreed measure for each
  • Regular reviews with leadership

Focused project

A defined problem, start to finish

Some businesses know exactly what they need built. We take that on with a clear scope, deliverables and success measures, and hand over something documented and owned by you.

Ongoing maintenance is available separately when the engagement is not a recurring one.

  • Fixed scope and deliverables
  • Agreed success measures
  • Maintenance quoted separately

Who does the work

The people in the discovery session are the people who write the code.

Obsidian Labs is a small studio. There is no layer between the person who understands your operation and the person who builds against it, because there is nobody to hand it to. That constraint is why we work the way we do — one improvement at a time, scoped to what a team this size can genuinely deliver and stand behind.

We spend the early part of an engagement in the work itself: the inbox, the dispatch board, the quoting spreadsheet, whatever actually runs the business. Most of what we learn there is not in anyone’s process documentation.

We build in your accounts, on conventional infrastructure, and we document as we go. If we stopped working together, everything would keep running and someone else could pick it up.

Founded
2023
Based
United States, working remotely
Clients
Established mid-market businesses, any industry
Engagement
Ongoing partnership, or a focused project
Ownership
Your repository, your infrastructure, your data

What could run better?

Tell us where work slows down, where visibility is missing, or where your team spends time it shouldn’t. We’ll explore what an improvement could look like.