MomentumOS

AI Transparency

Version 1.3 · Effective 11 July 2026 · Reviewed 6 November 2026

This page explains how MomentumOS uses AI to support executive intelligence while keeping deterministic platform calculations, evidence and human review at the centre of the experience.

Document control

Previous version: 1.2

Version: 1.3

Effective date: 11 July 2026

Last updated: 6 November 2026

Reviewed: 6 November 2026

Change summary: November 2026 cross-platform review: SourceMend data health, AI Connect MCP read-only access, tenant-gated Action Connect shadow proposals and approvals, independent ProProof execution boundary, updated UK privacy and cookies guidance, EU AI transparency, and clearer operational and commercial limitations.

What MomentumOS AI is used for

MomentumOS uses AI to explain business information, investigate performance changes, compare outcomes, generate executive summaries, prepare recommendations, draft action plans, prepare board briefings and summarise reports.

AI may help users understand Dashboard, Decision Room, Reports, Business Outcomes, Executive Intelligence Feed, Revenue Signals, Executive Risks, Executive Opportunities, Benchmarking, Workspace Health and Trust Centre content.

Ava and Ask MomentumOS use safe platform context, current workspace evidence and deterministic command routing where available.

What MomentumOS AI does not do

MomentumOS AI does not approve decisions, does not approve write-back, modify CRM records, modify PRM records, change integrations, approve production, access customer credentials or replace executive judgement.

AI outputs are not legal advice, financial advice, accounting advice, HR advice or guaranteed forecasting.

MomentumOS does not claim that AI-generated recommendations are final decisions. Humans remain responsible for reviewing and approving business actions.

Evidence, confidence and assumptions

MomentumOS presents AI-assisted outputs with evidence context where available, including source systems, data period, validation status, assumptions, missing evidence and confidence signals.

Deterministic MomentumOS calculations remain authoritative. AI may explain or summarise those results, but it does not replace the underlying calculation logic.

Where evidence is incomplete, MomentumOS should state that the evidence is not available yet rather than inventing metrics.

Private and ecosystem intelligence

Ava must distinguish private workspace observations, aggregate benchmarks, approved shared patterns, labelled inferences and labelled assumptions.

Ava must not reveal cohort members, private customer evidence, raw records, credentials, customer-specific embeddings, raw prompts or transcripts.

Shared ecosystem intelligence, where enabled, may only use approved, transformed, aggregated and de-identified observations from explicitly participating organisations.

Ava should describe ecosystem intelligence as readiness-gated unless production activation approvals are recorded.

Human review

Users should review recommendations, investigations, briefings, reports, action plans and scenario outputs before relying on them for business decisions. Approval boundaries are used for sensitive or operational actions.

Controlled actions and integration boundaries

AI may explain controlled outbound actions, webhooks, embeds and headless API status, but it does not enable or execute those capabilities by itself.

Outbound actions require permitted workflows, explicit customer approval where applicable, field allowlists, MomentumOS approval controls and audit records before execution.

AI prompts and responses are not intended to include customer credentials, OAuth tokens, API keys, client secrets, certificates, private keys or webhook signing secrets.

Persistent executive investigations and decision history

Where enabled, MomentumOS may persist executive investigations, evidence snapshots, confidence notes, assumptions, decision history, executive briefings and generated reports so authorised users can review how an output was prepared.

Voice and transcripts

Voice features support navigation, questions and briefings. Raw voice transcripts are not stored by default. If transcript storage or expanded AI memory is introduced later, it should be controlled by workspace settings and customer agreement.

AI Connect and the Model Context Protocol

AI Connect can supply tenant-scoped MomentumOS summaries, assessments and evidence to authorised third-party AI clients through MCP. It is read-only by default and uses separately scoped access. External AI providers' processing, memory and retention depend on those providers and customer account settings.

The MCP action:propose capability, where granted, supports drafting an internal proposal; it does not call an executor or bypass human approval. Users should verify which client receives the information before connecting.

Action Connect and independent assurance

Action Connect can convert eligible evidence-backed findings into proposed internal tasks, capture precise evidence and payload hashes, and require an authorised human approval. In the current tenant-controlled shadow workflow, approvals do not trigger downstream work.

Live action execution requires a distinct protected executor and an independent ProProof assurance decision, with evidence and policy binding, where configured and validated. A model response, permission to read data or recorded approval alone never amounts to admission to execute.

SourceMend and imperfect data

SourceMend checks uploaded or connected data for quality problems and can present proposed repairs for human review and export. It does not guarantee clean data or silently correct the customer's source system. Missing, stale, conflicting or incomplete evidence can limit or prevent recommendations and proposed actions.

Transparency, oversight and applicable AI rules

Users should be informed when they are directly interacting with AI, and machine-generated content should be identifiable where relevant rules require it. EU AI Act Article 50 transparency requirements began applying on 2 August 2026 to covered systems and uses, with specific transitional provisions; whether an obligation applies depends on the system, role, jurisdiction and use case.

MomentumOS is positioned as executive decision support, not an autonomous final decision-maker. Customers should independently assess their own AI deployment obligations, including appropriate human oversight, notices, fairness, record keeping and restrictions on prohibited or high-risk uses.

Outputs may contain errors or incomplete inferences and are not guaranteed forecasts. No claim is made here that MomentumOS or every connected model has been independently certified as compliant with the EU AI Act.