01
Human decisions remain human
ANCHOR can frame questions, identify gaps and signal review needs. The authorised person or group remains responsible for deciding what action is appropriate.
Responsible AI · Data · Governance
Decision support without automated judgement.
ANCHOR uses AI to help surface missing context, assumptions and unresolved risk in consequential people decisions — including decisions made by managers and by People teams themselves. It does not make employment decisions, replace professional judgement or remove organisational accountability.
Human accountablePeople remain responsible for the decision, review, advice and organisational action.
Advisory onlySignals prompt review; they do not create automatic verdicts or sanctions.
Fail closedIf analysis fails, ANCHOR does not fabricate a clean result or false certainty.
TraceableThe scenario, challenge, review and agreed route remain connected in one record.
The operating principles
The system is designed to improve the quality of human judgement — not replace it.
ANCHOR sits between a person’s first interpretation and the action that follows. That person may be a manager, HR professional, Employee Relations lead, People partner, senior leader or another authorised decision-maker. ANCHOR introduces structure, challenge and visibility while responsibility remains with the organisation.
02
No hidden automatic action
A signal cannot automatically begin a formal process, change employment status or issue an organisational judgement.
03
Evidence before labels
The workflow separates observed information from interpretation so assumptions are easier to identify and challenge.
04
Context is treated as essential
Health, disability, access, support, process and working conditions can be surfaced before a route hardens into action.
05
Uncertainty stays visible
ANCHOR does not present incomplete or failed analysis as a confident answer. Missing information remains explicit.
06
Review is proportionate
Not every decision requires escalation. Signals help the organisation direct review to the decisions where additional challenge, independence or specialist input is most useful.
Where AI sits
AI supports one part of a wider governed workflow.
The model is not the decision-maker. It contributes structured challenge inside a process that can be used by managers, People teams, Employee Relations, senior leaders and other authorised decision-makers, with review routed to the person or function best placed to provide appropriate independence and expertise.
01 · Human input
The decision-maker frames the situationObserved facts, interpretation and intended action are captured, whether the decision begins with a manager, People professional or another authorised role.02 · AI support
ANCHOR surfaces gapsMissing context, assumptions and review signals are presented as advisory prompts.03 · Human review
The appropriate reviewer respondsGuidance, challenge or confirmation is added by the person or function with the right authority, independence or specialist expertise.04 · Organisational action
An accountable route is agreedThe organisation decides what happens next and who owns it.05 · Traceability
The Decision Receipt remainsInputs, signals, response and outcome stay connected.Boundaries
What ANCHOR does — and what it deliberately does not do.
ANCHOR does not assume that risk sits only with managers. It examines the quality of the decision itself, whoever is making it, and routes review according to the organisation’s governance model.
ANCHOR does
- Structure a live people-decision scenario
- Separate observation from interpretation
- Surface missing context and unanswered questions
- Raise advisory or escalation signals for human review
- Connect organisational guidance and review to the original decision
- Create a usable decision record
- Support pattern review across repeated decisions
ANCHOR does not
- Decide whether an employee is right or wrong
- Automatically recommend dismissal or formal action
- Replace HR, Employee Relations, Occupational Health or legal advice
- Diagnose disability, neurodivergence or health conditions
- Guarantee that a decision is lawful or risk-free
- Hide uncertainty behind a fabricated score
- Remove accountability from managers, People teams, leaders or the organisation
Data lifecycle
Data handling is agreed as part of deployment — not assumed.
Before a pilot or licence begins, the organisation agrees what data is needed, who can access it, how long it is retained and how deletion or export will be handled.
Example governed data lifecycleConfigured per deployment
Capture
Minimum useful information
Only fields needed to structure, review and resolve the decision are agreed for the implementation.
Process
Structured analysis
Scenario information is used to generate advisory prompts and decision signals for authorised human review.
Access
Role-based visibility
Decision-maker, reviewer, governance and administrator access is defined according to the organisation’s operating and assurance model.
Retain or delete
Agreed lifecycle rules
Retention, export and deletion expectations are documented before live use and reviewed during the pilot.
Procurement note: exact hosting, processor, retention, access-control and contractual details should be confirmed in the implementation documentation supplied for the relevant pilot or licence. This page describes the product’s governance approach; it does not replace a completed security or data-protection review.
Access and accountability
Visibility should follow responsibility.
ANCHOR is designed so that access to decision information can reflect the organisation’s operating model rather than exposing every record to every user.
Role
Typical visibility
Typical responsibility
Decision-maker
The decisions they submit and the guidance connected to those decisions
Provide accurate context, distinguish evidence from interpretation, consider challenge and act within organisational guidance
Reviewer or specialist function
Decisions routed for review, unresolved signals and the context needed to assess them
Challenge, guide, confirm, escalate or seek additional expertise, then record the organisational response
Authorised governance lead
Approved pattern-level or implementation information
Oversee adoption, access, retention, assurance and improvement
Fail closed
The system should never invent certainty when analysis is unavailable.
A failed AI response is not converted into a false clean result, fabricated risk score or pretend Decision Receipt.
Possible system states
CleanNo specific review signal identified from the information provided.
AdvisoryQuestions or missing context should be considered before proceeding.
EscalateHuman review is recommended before the intended action proceeds.
Failure is shown as failure.
If the AI service cannot return a valid analysis, ANCHOR should stop that part of the workflow and make the failure visible.
It should not silently assign a lower-risk result, create a completed receipt or allow a technical error to appear as organisational approval.
The next action is operational: retry, seek human review or use the organisation’s fallback route.
Before live use
A pilot includes governance decisions as well as product testing.
The implementation should establish the operating conditions around ANCHOR before real organisational decisions are processed.
Purpose and scope
Which decision types are in scope, which are excluded and what the pilot is intended to learn.
Data fields
What information is genuinely needed and what should not be entered into the system.
User roles
Who can submit, review, administer, export and see pattern-level information.
Review thresholds
How advisory and escalation signals connect to the organisation’s existing routes.
Retention and deletion
How long pilot records are needed and what happens at the end of the engagement.
Incident and fallback route
What users do if analysis fails, information is wrong or urgent human review is required.
User communication
How all users — including managers, People professionals and leaders — are told what ANCHOR does, what it does not do and how their inputs will be used.
Success and assurance review
How the organisation will assess usefulness, proportionality, user behaviour and governance friction.
Enterprise questions
What procurement, legal, IT, People teams and organisational leaders usually need to know.
Does ANCHOR make employment decisions?
No. It provides structured prompts and advisory signals. The organisation remains responsible for review, advice and action.
Is a clean signal the same as approval?
No. It means no specific signal was identified from the information provided. It is not legal approval, policy approval or a guarantee that the decision is appropriate.
Can ANCHOR replace professional or independent review?
No. ANCHOR can support managers, People teams and other decision-makers, but it does not replace professional judgement, specialist advice or an appropriately independent review route.
Can ANCHOR review decisions made by the People team?
Yes. ANCHOR is designed for consequential people decisions regardless of whether they originate with a line manager, HR Business Partner, Employee Relations lead, People Director, senior leader or another authorised role. Review should be routed to someone with appropriate authority, independence or specialist expertise.
What data is required?
The minimum useful fields should be agreed for the deployment. Users should not enter information that is unnecessary for the decision workflow.
Who can access decision records?
Access should be configured according to role and organisational responsibility. The intended model is not universal visibility.
What happens if the AI output is wrong or incomplete?
The output is advisory and remains subject to human review. Users should be able to challenge, correct or disregard it, and technical failures should remain visible rather than being converted into a false result.
Does this page replace a security or data-protection review?
No. Detailed assurance should be completed against the actual deployment, contracts, processors, retention rules and access model agreed with the organisation.
Assurance before adoption
Map ANCHOR into your governance environment before it goes live.
Bring your People, leadership, data, security, legal or procurement questions. We will separate fixed product principles from the controls that need to be agreed for your organisation.
