CortexForge unveils Invisible Intelligence for Microsoft

CortexForge has introduced Invisible Intelligence, a framework for coordinating AI across organisations. It is designed to keep information moving within existing Microsoft environments under defined access controls.

The Saskatoon-based consultancy says the model is intended to reduce the manual work involved when a single incident must be repeatedly summarised, redacted, forwarded and re-entered across departments.

Invisible Intelligence is built on the idea that an event should be recorded once, then interpreted against predefined organisational roles, permissions, escalation rules and approved data sources. The goal is to prepare different views of the same event for authorised users without giving every stakeholder access to the full underlying record.

In one example, a campus security incident involving a fight and property damage could generate separate updates for facilities staff, student support teams, counselling professionals, security leaders and senior executives. Each group would receive information relevant to its role, while personal details and broader records would remain restricted unless specifically authorised.

The approach reflects growing pressure on organisations to use AI tools without weakening privacy controls or handing decision-making to automated systems. CortexForge says the framework keeps consequential decisions, including disciplinary action and care-related judgments, with human staff.

Microsoft stack

The architecture sits within Microsoft software already used by many organisations. Microsoft Dataverse serves as the governed operational record, Microsoft Copilot Studio handles the intelligence and agent layer, and Microsoft Power Automate manages workflow execution.

Microsoft Entra defines an agent's identity and access boundaries. In practice, that means an AI agent assigned to one task, such as facilities coordination, would receive only the access needed for that purpose, not unrelated records such as counselling information.

CortexForge describes the structure as a separation between intelligence and execution. Under this model, the system may identify that a threshold has been met, prepare the relevant information and begin an approved workflow, but it does not have open-ended authority to decide an organisation's response.

This distinction is central to the framework. CortexForge argues that many businesses do not lack information; they struggle to move it to the right people quickly enough and with the right level of detail.

Governance focus

CortexForge says it developed the governance model for Invisible Intelligence through its Ethics Division. The framework sets out who acts first, who verifies activity, what information may be accessed, when automation stops and where accountability rests.

The objective, according to the consultancy, is to prevent responsibility from becoming blurred when AI systems are introduced into routine operational processes. Recommendations, notifications, escalations and approvals can remain tied to the underlying operational record, allowing organisations to reconstruct what happened and why.

That emphasis on traceability comes as companies face scrutiny over how AI systems handle confidential information and support workplace decisions. Tools that route or summarise data across departments can improve efficiency, but they also raise questions about over-disclosure, unauthorised access and whether staff understand how the systems operate.

Invisible Intelligence is intended to address those concerns by preparing separate views of the same event according to role and purpose, rather than producing one general summary for wide distribution. Facilities teams might see only repair and location details, while senior leaders could receive a de-identified update focused on operational impact and urgency.

"The more invisible the intelligence, the more visible its boundaries must be. By placing automation inside the organization's own infrastructure, work can move quietly in the background without authority disappearing into the system. Intelligence orchestrates. People remain in command," said Maverick Tarasoff, President, CortexForge.

Frontline trust is another issue in AI adoption. Staff are less likely to rely on a system if they do not know what it can access, what actions it is taking in the background or how its outputs may affect colleagues, students or customers.

"Frontline workers will not use AI they do not trust. That trust comes from transparent ethics. So teams can work alongside intelligence with confidence," said Shayla Tarasoff, Vice President, CortexForge.

Invisible Intelligence is the latest part of CortexForge's broader work on client-owned operational systems. Its wider approach centres on keeping records, workflows, permissions and data inside environments governed by the client rather than external platforms, while using AI to coordinate information without removing human authority.