Skip to content

Product 01

Active · Activation stage

Compliance Intelligence Platform

CIP assesses how well organisations meet their obligations — and shows the evidence for every judgement it makes. It is the first major platform powered by CIM v1.0, Sentient's Compliance Intelligence Model.

Request a demonstrationExplore the interface
CIP · Compliance Intelligence Platform
Criteria defined in the framework
132
Weighted categories
13
Determination reason codes
40
Automated tests
492

01The problem

Knowing whether obligations are met has meant months of manual review.

Paper returns. Scattered websites. Policies that change without notice. Findings that cannot be traced back to anything. When compliance is assessed by hand, it is slow, inconsistent and difficult to defend — and it rarely covers everyone it should.

02The solution

Compliance assessment you can audit, not just read.

CIP reads the evidence, checks it against every defined obligation, and turns it into scores, grades and ranked advice — consistently, at scale, with the source and the exact words behind every finding.

  • Principle 01

    A score is a consequence of evidence.

    Never something typed in. Every determination carries its source, the page or clause that decided it, the words that were read and a written reason.

  • Principle 02

    If it could not look, it says so.

    Where CIP cannot establish something, it records it as unknown and names the shortfall — rather than scoring a failure.

  • Principle 03

    CIP proposes. People decide.

    Nothing it concludes is acted on without a person. Every decision that can move a score is recorded against a name and sealed.

03How CIP works

CIP does not make legal determinations. It analyses evidence against criteria an organisation has defined, and produces structured compliance intelligence for people to review.

PASSPARTIALFAILUNKNOWN→ FOR HUMAN REVIEW
01Data
  1. 01

    Data

    Evidence is gathered from every available source — registers, filed returns, published pages and measured signals — and ranked by authority.

  2. 02

    Criteria

    Each obligation is a defined criterion: the phrases and patterns that evidence it, its weight, and the standard of search required before an absence can count.

  3. 03

    Processing

    Deterministic matching reads the evidence first and keeps the exact words that produced each result. AI-assisted reading is consulted only where matching cannot decide.

  4. 04

    Analysis

    Every criterion receives a status — pass, partial, fail, not applicable or unknown. An absence becomes a failure only when the search genuinely met the evidence standard.

  5. 05

    Intelligence

    Determinations become scores, grades and ranked advice — computed from the evidence on read, so a figure can never disagree with what lies beneath it. Thin evidence withholds a grade; it never becomes a zero.

  6. 06

    Compliance output

    Findings, evidence and reasons are presented to people. They review, decide and act — and every decision is written to a sealed record.

04The model

The Compliance Intelligence Model is Sentient's first deployed intelligence model. It analyses compliance against defined laws, acts, regulations, obligations, policies, criteria, metrics and organisational requirements — and supports monitoring, analysis and decision-making. It does not replace legal judgement.

Evidence is gathered from every available source — registers, filed returns, published pages and measured signals — and ranked by authority.

Sources, in order of authority

  1. 00A reviewer's determinationA named person's judgement outranks every automated source.
  2. 01Official registersAuthoritative records held by the responsible body.
  3. 02Filed returnsCompliance returns and documents, read clause by clause.
  4. 03Published materialWhat an organisation publicly serves — read responsibly, cited verbatim.
  5. 04Measured signalsSecurity and configuration properties observed, not assumed.
  6. 05AI-assisted readingConsulted only where deterministic analysis cannot decide.

Designed to help organisations understand

  • Are required criteria being met?Each criterion receives a status, with the evidence that decided it.
  • Where are the gaps?Unmet and partly met obligations are identified, with the reason written down.
  • What needs attention first?Advice is ranked, so the most consequential gaps come first.
  • How are we performing?Scores and grades by category, against defined requirements.
  • What patterns are emerging?Compliance across many organisations and large datasets, seen together.

Every criterion receives a status

  • Pass
  • Partial
  • Fail
  • Not applicable
  • Unknown
  • Laws
  • Acts
  • Regulations
  • Obligations
  • Policies
  • Compliance criteria
  • Metrics
  • Organisational requirements

05Inside the platform

  1. 01

    Data ingestion

    CIP draws evidence from registers, filed compliance returns, published material and measured signals. Each source has a rank, and a stronger source is never overwritten by a weaker one. A reviewer's determination outranks them all.

  2. 02

    Compliance criteria

    Laws, acts, regulations, obligations and policies become a framework of criteria held as data — weighted by category, each with the phrases that evidence it and the standard of search it requires. The framework changes only through an approved change set.

  3. 03

    Compliance monitoring

    Assessment runs work across every organisation in scope. Watched policy pages are compared over time, and a substantive change raises an alert.

  4. 04

    Intelligent analysis

    CIM v1.0 analyses the evidence against each criterion and keeps the verbatim words that decided it. Filed returns are read clause by clause — distinguishing an assertion from a denial, a plan or a quotation of the law.

  5. 05

    Compliance outputs

    Statuses, scores, grades and ranked advice — each traceable to its evidence. Where too little could be established, the grade is withheld with the reason stated.

06Surfaces

Dashboards, scores, evidence, advice and alerts — each one computed from the evidence beneath it, so nothing on a screen can disagree with what it rests on.

  • Dashboards

    The overall picture at a glance — what is graded, what needs more information, and why. Every figure is computed from evidence when it is read.

  • Scores & analytics

    Scores broken down by weighted category, rescaled to what could actually be assessed, with grade bands from A to E.

  • Evidence register

    Every determination points to a record: the page that was read, the words quoted from it and how it was obtained. Replaced evidence is kept as history.

  • Advice & reports

    Ranked remediation advice per organisation, and exportable decision records — so a finding can be acted on, not merely read.

  • Alerts

    Watched policy pages raise an alert when they are published, edited or taken down — compared on their substance, not their menus.

  • Directory & profiles

    Every organisation assessed, filterable by grade and status, each opening a profile with every obligation, its evidence and its reason.

07Workflow

  1. 01

    Assess

    An administrator starts an assessment run. It can be stopped at once — then continued, restarted or cancelled.

  2. 02

    Supply

    Staff add evidence a crawler cannot see: filed returns, addresses, a reviewer's own determination — which outranks automated sources.

  3. 03

    Review

    People read every finding with its evidence and reason, and the advice ranked behind it.

  4. 04

    Decide

    Decisions are taken by people and sealed into the record, readable by the person who made them.

  5. 05

    Improve

    Calibration tests proposed changes against examples people have already judged — accuracy before and after — prior to any approval.

08Architecture

A model, an evidence store that never forgets what it once believed, a scoring engine that computes on read, and a sealed record of every human decision.

Sourcesranked by authority
  • Reviewer's determination
  • Official registers
  • Filed returns
  • Published material
  • Measured signals
CIM v1.0the model
  • Deterministic matching
  • Clause-level document reading
  • AI-assisted reading (only where needed)
Evidence storecurrent + history
  • Determinations
  • Verbatim evidence
  • Score snapshots
  • Replaced evidence kept as history
Consolecomputed on read
  • Dashboard
  • Directory & profiles
  • Scores
  • Evidence register
  • Advice
  • Alerts
Decision record

Append-only. Every decision a person makes is recorded against their name, each entry sealed to the one before it.

Maintenance & testing

Engineers change the framework only in a testing environment, through change sets an administrator approves before anything goes live.

Architecture at a glance · local deployment · nothing leaves the machine

09Use cases

  • 01

    Assess compliance at scale

    Measure many organisations against the same defined obligations, consistently, with the evidence for every finding attached.

  • 02

    Prioritise enforcement

    Inform proportionate responses where an organisation defaults — grounded in recorded evidence, decided by people.

  • 03

    Direct investigations

    See where gaps cluster and which organisations need attention first.

  • 04

    Understand a whole sector

    Research the state of compliance across an entire landscape — and see honestly where evidence is still missing.

Deployment environment

CIP runs locally, on the organisation's own machine — no external database server, and nothing leaves it. Access is controlled entirely by the organisation.

10Safeguards

No single person can both change how CIP judges and let that change into the live platform.

  • Human oversight

    CIP proposes; people decide. Decisions that can move a score, remove evidence or change what is measured are recorded against a name.

  • Sealed decision record

    An append-only audit log where each entry is sealed to the one before it, so an edit or deletion is detected.

  • Access by token

    Each person holds a token issued for a role and a period — Viewer, Operator, Administrator or Maintenance Engineer — revocable at any moment.

  • Separation of duties

    Engineers change what is measured only in a testing environment; nothing goes live until an administrator who did not make it approves it.

  • Responsible reading

    CIP identifies itself, honours robots.txt, uses read-only requests and never logs in, submits, probes or tests anything it reads.

  • Data stays local

    The platform runs on the organisation's own machine. No external database server; nothing leaves it.

11Demonstrations

A designed demonstration of CIP's interface and flow. Live demonstrations of the platform are available on request.

Evidence arrives from every available source — ranked by authority.

Interactive — choose a view

We will walk you through the platform, the model and the safeguards — and how CIP could apply to your obligations.

Request a demonstration