AI for Research
Integrate AI into research work transparently, with source discipline, institution-specific rules, documentation, verification, and a real human review gate.
Request a fit callRecommended starting point
A 90-minute online review plus a short workflow and next-step plan. We map where AI may help, where it should not be trusted, and how every critical output will be checked.
You receive the complete format and written plan. In return, I ask for structured private feedback that helps refine the service; public feedback is optional.
Read the session, payment and cancellation arrangements
Illustrative agenda, adapted by agreement to your task. This is a planning example, not a guarantee that a project is solved in one session.
Bring an anonymised or synthetic example and the rules that apply. The agreed written plan records priorities and checks; ongoing implementation or unlimited support is not included unless separately agreed.
Inspect a sample plan and checklistInteractive process
Explore the stages and a worked example. Hover or focus to preview; click or tap to keep a stage selected.
We define the question and its boundaries before assigning tasks to AI. The question determines which evidence and methods count.
Record the question, search terms, inclusion criteria, and the limits of the overview.
Search and triage focus on real, verifiable sources; model output is not a source. Traceability and source-quality decisions begin here.
Capture real titles, links or identifiers, study designs, and reasons for inclusion.
We compare evidence, methods, and disagreements while keeping every claim tied to real sources. AI may organise ideas, but it cannot confer authority.
For each claim, record the source, method, finding, and limitation. Keep disagreements visible.
Design, assumptions, checks, and decision points are specified before analysis or coding. This makes AI assistance testable, explainable, and easier to debug.
Decide how extraction will be checked and how uncertain or conflicting results will be handled.
AI is used selectively for structure, clarity, or iteration while human authorship remains intact. Its use is documented and disclosed according to institutional rules.
Use AI only within the agreed rules. Keep source links and document how assistance was used.
Every critical claim, source, calculation, code result, and disclosure passes a genuine human check. This is a mandatory gate, not a decorative ending.
Open the original papers. Check claims and references. Resolve unsupported statements before use.
We define the question and its boundaries before assigning tasks to AI. The question determines which evidence and methods count.
Illustrative example: Record the question, search terms, inclusion criteria, and the limits of the overview.
Result: A scoped research aim
Search and triage focus on real, verifiable sources; model output is not a source. Traceability and source-quality decisions begin here.
Illustrative example: Capture real titles, links or identifiers, study designs, and reasons for inclusion.
Result: A relevant, verifiable source base
We compare evidence, methods, and disagreements while keeping every claim tied to real sources. AI may organise ideas, but it cannot confer authority.
Illustrative example: For each claim, record the source, method, finding, and limitation. Keep disagreements visible.
Result: A traceable evidence map
Design, assumptions, checks, and decision points are specified before analysis or coding. This makes AI assistance testable, explainable, and easier to debug.
Illustrative example: Decide how extraction will be checked and how uncertain or conflicting results will be handled.
Result: A defensible analysis plan
AI is used selectively for structure, clarity, or iteration while human authorship remains intact. Its use is documented and disclosed according to institutional rules.
Illustrative example: Use AI only within the agreed rules. Keep source links and document how assistance was used.
Result: A transparent writing workflow
Every critical claim, source, calculation, code result, and disclosure passes a genuine human check. This is a mandatory gate, not a decorative ending.
Illustrative example: Open the original papers. Check claims and references. Resolve unsupported statements before use.
Result: A reviewed, documented result
What we work on
Models can support brainstorming, source triage, synthesis, note organization, coding, debugging, analysis planning, and writing. They are not authoritative sources and they do not remove the need to inspect evidence.
Plan searches, triage materials, organize notes, compare arguments, and keep every claim traceable to real sources.
Use AI for planning or debugging while documenting choices, testing output, and understanding assumptions and failure modes.
Create transparent support routines that respect university policies, citation rules, disclosure requirements, and human authorship.
AI Act or organizational compliance questions receive general orientation only. Legal questions belong with qualified legal or compliance professionals.
Academic integrity
Support is educational, methodological and editorial. It does not replace your authorship, judgment, or assessed performance.
Questions before you decide
No. Models can help organize or question material, but claims must be checked against real, appropriate sources.
Yes, as a transparent support workflow: document choices, test outputs, understand assumptions, and retain human responsibility for interpretation.
No universal guarantee is possible. We work from the rules that apply to your institution, project, data, and discipline.
No. Practical workflow implications can be discussed, but legal and formal compliance decisions belong with qualified specialists.
A grounded next step
The short intake identifies the aim, constraints, and right format before any paid work begins.
