Tenzing.

    Agent Governance Framework

    Has governance kept pace with your AI ambitions?

    Copilot is scaling faster than the governance to control them. Most organisations don't realise the gap until something goes wrong.

    Book a 30-minute conversationSee the framework →

    AI moves faster than governance can keep up.

    Agents are in production that nobody can fully account for.

    Built by business users, approved informally, running on identities and permissions nobody documented.

    Sensitive data is moving through systems with no audit trail.

    Not because anyone meant for it to, but because the boundaries weren't set before the agents were built.

    The conversations are getting harder.

    Security wants answers nobody has. Audit wants evidence nobody assembled. The board wants reassurance the team can't yet give.

    The Solution

    A practical framework for governing Microsoft agents.

    01

    Outcomes you can point to.

    Defines what good looks like in business terms. Programmes that ship, agents that work as intended, evidence that exists when asked for.

    02

    Built for the platform you actually use.

    Specific to Microsoft 365, Copilot, Power Platform, Purview and Azure AI Foundry. The platform your business already runs on.

    03

    A maturity path, not a checklist.

    Three tiers that meet you where you are. A defensible foundation first, a credible path to scale next.

    What good looks like at each level.


    Three levels of maturity, defined by outcomes rather than controls. The reference points to test against at each tier.

    MVS
    Minimum Viable Security
    Before any further agent deployment

    Clear ownership and basic guardrails in place.

    The platform is safe by default.

    Every agent is known, reviewed and stoppable.

    RECOMMENDED
    Recommended Controls
    MVS complete · Months 1 to 12

    Governance operates across the business.

    Environments, data and identity are managed at scale.

    Agents are engineered and incidents are owned.

    ENTERPRISE-READY
    Advanced Controls
    Recommended embedded · 12 to 24 months

    AI governance is embedded in operations.

    The platform self-corrects across the estate.

    Agent development is industrialised.

    Benchmark your position. Plan your next move.

    A fixed-scope, fixed-price engagement that takes you from "what good looks like generally" to "what good looks like for us, and what to do first".

    01

    Honest baseline.

    Where you actually sit against MVS, Recommended and Enterprise-Ready. With evidence, not aspiration.

    02

    Prioritised gaps.

    What's missing at each level, weighted by severity and operational impact. Specific to your platforms and posture.

    03

    Actionable roadmap.

    A phased plan tied to your risk appetite, your platforms and your timeline. Sequenced for your team to actually deliver.

    What Gets Delivered

    Five stages. Two weeks. One actionable report.

    From objectives to roadmap, structured to deliver clarity your stakeholders can act on.

    01

    Objectives

    What success looks like, business context, platforms in scope, stakeholders.

    02

    Current state

    Discovery across people, process and platform. Interviews, document review, working session.

    03

    Framework benchmark

    Position against MVS, Recommended and Enterprise-Ready. Visualised, not numeric.

    04

    Gap analysis

    What's missing at each level, weighted by severity and operational impact.

    05

    Roadmap

    Phased plan tied to MVS, then Recommended, then Enterprise-Ready. Sequenced, scoped, realistic.

    Ready to Start?

    See where you stand against the framework.


    Twenty minutes to compare your current state against the framework. We share what we see most often in your sector. You share what you're trying to solve. No prep required.

    No sales pitch · No commitment · 30 minutes

    Eoin Fahy · Founder, Tenzing · eoin@tenzing.it.com