Checkpoint Packs
One judgement layer. Different professional contexts.
Unthinkit Beta provides general coaching questions for AI-generated answers. Specialist packs are future functionality: domain-specific checkpoints for the assumptions, evidence and professional decisions that matter in a particular field.
Development principle
Specialist packs should be built with specialists.
A domain pack should not simply add more generic prompts. It should reflect the professional evidence, standards, terminology and judgement points that practitioners actually use.
Specialist packs will therefore be developed and tested with relevant domain experts before release. The aim is to make the checkpoint library more professionally useful without turning Unthinkit into a substitute for professional responsibility.
Roadmap
From easier-to-validate domains to higher-stakes ones.
The order below is a development direction, not a release commitment. Domains with greater safety, regulatory or professional complexity require correspondingly stronger validation before they should become specialist packs.
| Domain | General Unthinkit can already help examine | Specialist pack could add | Status |
|---|---|---|---|
Valuation & Property | General coaching questions about passages that may deserve another look. | Comparable evidence, valuation assumptions, adjustments, sensitivity, market context, policy and professional standards. | Next |
Public Sector | General coaching questions about passages that may deserve another look. | Statutory duties, policy authority, equality considerations, procedural context and accountable decision records. | Planned |
Business | General coaching questions about passages that may deserve another look. | Unit economics, market-size extrapolation, incentives, scenario dependencies and commercial execution risk. | Planned |
Recruitment | General coaching questions about passages that may deserve another look. | Job-related evidence, consistency, decision rationale and domain-specific fairness checkpoints. | Planned |
Audit & Accounting | General coaching questions about passages that may deserve another look. | Materiality, controls, accounting standards, audit assertions and evidence sufficiency. | Planned |
Investment | General coaching questions about passages that may deserve another look. | Sensitivity, scenario assumptions, valuation inputs, concentration and risk-return dependencies. | Planned |
Law | General coaching questions about passages that may deserve another look. | Authority hierarchy, jurisdiction, precedent relevance, interpretation and procedural context. | Planned |
Science & Research | General coaching questions about passages that may deserve another look. | Sample design, controls, statistical assumptions, evidence hierarchy and transferability boundaries. | Planned |
Code & Systems | General coaching questions about passages that may deserve another look. | Edge cases, dependencies, testing, security boundaries and destructive-operation risk. | Planned |
Clinical / Health | General language-level checkpoints only. Unthinkit 1.0 is not a clinical decision-support tool. | Any specialist health pack would require development and validation with appropriately qualified clinical experts, clear scope boundaries, safety review and consideration of applicable regulation before release. | Exploratory only |
Why valuation comes next
Professional judgement is already visible in the workflow.
Valuation is a natural first specialist domain because automated drafting and analysis can assist the work while the professional still has to weigh evidence, assumptions, comparables, adjustments, uncertainty and the final conclusion. A Thinking Receipt can record where that judgement entered the process without claiming to certify the valuation itself.
Important boundary
A specialist pack supports judgement. It does not replace the specialist.
A Thinking Receipt records review activity; it is not professional sign-off, certification or proof that the underlying output is correct.
This boundary is especially important in health and other safety-critical domains. No clinical or health specialist pack will be released merely by adapting the general checkpoint library. Any such work would require appropriately qualified clinical collaboration, safety and regulatory review, explicit intended-use boundaries and validation appropriate to the proposed use.