Methodfield is a case about how I connect management methodology, process design and the creation of a working digital product.
The starting problem is to help a person move from an unclear work issue to a suitable method and a concrete action. The project connects an open library, explainable method selection, practical materials and saved working plans.
The case presents three key decisions:
- Explainable method selection: recommendations account for context, available evidence and constraints through rules that can be examined.
- A path to action: users record facts, a hypothesis, a next step, an owner and a review date.
- System reliability: saved data is supported by access isolation, protection against invalid changes and restoration checks.
My contribution is defining the purpose, methodological principles, product structure and outcome requirements, and directing development with AI tools. I pay particular attention to coherence across content, the user journey and technical implementation.
The result is a live platform where people can examine my approach and check specific features. The case demonstrates a connection between analysis, implementation and quality control. The platform's effect on learning and users' work outcomes remains the next subject for investigation.
The problem I chose to address
Knowing the names of management methods does not establish which one fits a particular situation. A team investigating repeated errors needs to distinguish a symptom, a causal hypothesis and evidence that would support a change. A team choosing between options needs explicit criteria and an understanding of the consequences of being wrong.
My product hypothesis was that a connected path could make this reasoning easier to apply: clarify the problem, select a method, understand its limits, define an action and return to the result. Methodfield implements parts of this path that people can inspect and use. Its effect on learning and workplace outcomes still needs separate measurement.
Decision 1: make method selection explainable
The Method Finder considers eight structured dimensions, including the desired outcome, available evidence, consequences of error, working time and reversibility. Its ranking follows explicit rules. Recommendations explain suitability, cautions and supporting methods that may belong before, alongside or after the primary method.
I wanted the selection logic to remain open to examination. A free-form model recommendation would introduce variability and make it harder to trace a result to the product's methodological rules. The implemented rule-based approach makes its assumptions and behaviour more explicit.
That choice has a maintenance cost: rules must evolve with the catalogue and remain consistent across languages. It also has a clear limit. A fit score describes compatibility with stated criteria; it does not establish the probability that applying a method will solve the problem.
The professional judgement lies in defining the dimensions, distinguishing methods and preserving those limits. The software makes that judgement repeatable and inspectable.
Decision 2: connect a method to an owned action
Reading a guide leaves an important question open: what will the person actually do next?
Methodfield now connects method pages and Finder results to a working plan. The plan records the problem, known facts, a hypothesis, an action, an owner and a review date. It also includes a baseline and expected and observed results. Plans can be saved, reopened and exported; completion requires a recorded result.
This puts the handoff from explanation to action into the product itself. It also exposes a deliberate scope limit: the current plan is a common structure for applying a method, rather than a specialised interactive editor for every tool in the library.
This diagram describes the intended working sequence supported by the relevant features. It is not evidence that users consistently complete the sequence or improve their business outcomes.
Decision 3: make reliability part of the result
A saved plan becomes useful only if the user can return to it and trust how it is handled. Persistence introduced requirements beyond displaying a form: access isolation, input validation, protection against stale updates and predictable behaviour when an operation is retried.
The September release added those controls to the working-plan flow. Its validation also covered reopening a plan, recording a result and exporting the saved information. Operational work added notifications, external encrypted backups and a tested restoration procedure.
These are concrete expressions of a broader principle in the project: define what the system may do, how an outcome is checked and how work can recover after failure. The same reasoning informs Methodfield's material about AI authority and human control.
My role and the use of AI
My role is the author of the approach and the person directing the product's creation: defining its purpose, connecting its parts, setting constraints and taking responsibility for substantial decisions.
AI tools assist research, drafting and implementation. I describe their contribution alongside my own. Responsibility for a project should be visible through decisions, review and correction; it should not be confused with personally typing every line of code or executing every test.
The most demanding part has been keeping the methodology, content, user journey and technical behaviour coherent as the project grows. The September improvements are one example of addressing a gap between available knowledge and a user's ability to preserve a next step.
What the evidence supports
The public application lets a visitor inspect guides and the Finder. The release verification of 4–5 September 2026 records checks of plan creation, saving, reopening, access isolation, stale-update handling and exports. The validation included isolated restored databases and browser checks. A separate operations report records restoration from an external encrypted backup.
These are dated release results, not claims that every test was rerun for this article. They support the existence and tested behaviour of specific product features. They do not establish improved decision quality, learning retention or productivity across an external user population.
The next useful evidence is observation of people using this path: whether they understand the recommendation, formulate an actionable plan and return to evaluate what happened. Methodfield provides a working basis for that investigation.