Researcher resources
These resources are designed to be copied and adapted to another research project.
If you do not know where to begin, the decision guide connects common research problems to specific records. The ten-stage crosswalk shows where a record might help and where verification must occur independently of documentation.
They are deliberately lightweight. The aim is to record consequential decisions and claim boundaries, not every prompt, keystroke or conversational exchange. For a first session, begin with one safe task and one checked suggestion; you may not need a form at all.
Starting a project
AI-Assisted Research Charter
Use before substantial AI-assisted exploration to state:
- the substantive research problem;
- the analytical object;
- non-negotiable constraints;
- permitted AI roles;
- verification boundaries;
- consequential decisions that should be recorded.
Recording consequential decisions
AI-Assisted Research Decision Log
Use when a decision materially changes the analytical object, scope, evidence base, interpretation or next stage of the research. Keep the initial entry unchanged; append dated later updates. A retrospective reconstruction must be labelled as such.
Reason-Provenance Record
Use when a later contribution or attribution claim may depend on reconstructing how a consequential move entered the research process.
Researcher–AI Contribution Map
Use to distinguish origination, evaluation, integration and interpretation across selected research episodes.
Challenging claims
Claim Challenge Record
Use when a comparator, robustness check, reviewer, researcher or AI-assisted critique challenges a preferred claim.
Inference Revision Record
Use when a result remains numerically or descriptively correct but its substantive interpretation changes, narrows or is withdrawn.
Governing manuscript revision
Claim-Preservation Contract
Use once the substantive argument is sufficiently mature to state:
- authorised claims;
- prohibited claims;
- inferential boundaries;
- required qualifications;
- frozen conceptual or terminological distinctions.
Evidence Ledger
Use to connect substantive manuscript propositions to the evidence or reasoning that licenses them.
Post-Revision Sufficiency Audit
Use after substantial compression, restructuring or AI-assisted rewriting to check that the revised manuscript still contains the evidence, qualifications and distinctions needed to support the authorised argument.
Preparing material for release
Public Release Screening Checklist
Use before publishing process evidence, worked examples, slides or archival material.
Choosing proportionate records
You do not need every resource for every project. In an initial low-risk exchange with no consequential decision, no formal template may be necessary. If a suggestion changes the research, a dated decision note may suffice. Check the actual evidence independently of the notes.
One possible sequence for a complex project, where each step serves a real purpose, is:
Research Charter → selective Decision Log → Claim Challenge Record when needed → Claim-Preservation Contract during rewriting → Release Screening Checklist before publication.
Use provenance, inference-revision and evidence-ledger records when there is a specific attribution, interpretation or evidential question to resolve. Neither the number of forms nor their completeness proves research quality.
See the resources completed
A completed PISA example set shows how the records can be used in one research process.
The completed set is an exemplar rather than a required format for other projects.