Field notes
Lessons discovered inside organisations.
- Where a stated requirement differs from the real one
- How workarounds reveal the actual constraint
- What an operation cannot afford to have disturbed
- Observation
- Pattern
- Consequence
- Note
Research
Research for systems that must work in the real world.
We study new technology through the problems faced by actual organisations.
Our research covers AI systems, enterprise workflows, maritime operations, agent design, adoption, permissions, and the future of engineering.
We do not research technology because it is new. We research it to understand where it becomes useful.
Engineering notes
How systems are designed, evaluated, deployed, and operated.
- Defining responsibilities and limits for a governed AI worker
- Evaluating retrieval quality against real operational documents
- Introducing a system into a workflow that cannot stop
- Problem
- Hypothesis
- Experiment
- Finding
Maritime research
Technology, workflows, compliance, and operational change in shipping.
- Certificate and documentation workflows
- Compliance work across vessels and shore teams
- Modernising software that an operation still depends on
- Problem
- Hypothesis
- Experiment
- Finding
Frameworks
Methods for understanding workflows and introducing AI safely.
- Mapping a workflow before choosing a technology
- Deciding when not to use AI
- Earning autonomy gradually: retrieve, prepare, review, execute
- Question
- Method
- Application
- Revision
Experiments
Ideas we are testing before they become products or deployments.
- Natural language as the interface to a multi-step process
- Structuring organisational knowledge so a system can use it
- Problem
- Hypothesis
- Experiment
- Finding
Publishing
Nothing is published yet.
Forthcoming
We would rather publish nothing than publish a finding we cannot stand behind. Our first notes will come from work already under way: how a governed AI worker is given responsibilities and limits, how retrieval is evaluated against real operational documents, and what we have learned introducing a system into a workflow that cannot stop.
Each published note will carry its category, date, author, and status, and will follow the same structure we use internally: problem, hypothesis, experiment, finding.
We research before we recommend.
Before we recommend an intervention, we study the organisation, its industry, existing products, modern models, proven engineering methods, adoption risks, and long-term trade-offs.
We do not build from excitement alone. We build from evidence.