AI readiness on Microsoft 365: finish the intranet on G5 first
There is a familiar sequence in Microsoft 365 programs right now. Leadership asks about AI. Someone demos a Copilot-era feature. The intranet team is told to “add AI” to the homepage. Meanwhile, people still cannot find last year’s policy, permissions still leak titles through navigation, and orphaned team sites still clutter search.
That sequence has the work backward.
AI readiness on Microsoft 365 is mostly intranet readiness. If the access model is unclear and the information architecture is noisy, AI will summarize the noise faster. If the base intranet is complete on the Microsoft 365 E5 / G5 capabilities you already license, optional AI features become a controlled choice—not a rescue mission.
This is not “AI will replace your intranet team.” It is the opposite stewardship: finish the foundation people already pay for, then turn AI on when the organization can supervise it.
What “complete on existing G5” means
Many organizations already hold Microsoft 365 E5 or G5 (or equivalent bundles with advanced compliance, identity, and security). “Complete” here does not mean every feature toggle is on. It means the intranet job is done well enough that optional AI has something trustworthy to stand on:
Access model — Entra groups, RBAC, permission-respecting navigation (no greyed-out title leaks as a habit).
Information architecture — hubs with purpose, plain-language navigation, owned content.
Findability — search that returns the current policy, not three near-duplicates.
Lifecycle — owners, review dates, and a path to retire abandoned sites.
Sensitivity awareness — labels, retention, or DLP patterns appropriate to your risk—not theater checklists.
Those five are ordinary Microsoft 365 work. They are also the precondition for AI features that read across SharePoint, OneDrive, and Teams. If AI can see it, your access model already decided whether that is safe.
Optional and reversible—on purpose
Treat Copilot-era and similar AI capabilities as optional and reversible:
Optional — the intranet succeeds without them. AI is an accelerator for people who already know where truth lives, not a substitute for structure.
Reversible — you can narrow scope, pause a feature, or restrict audiences without redesigning the whole tenant. Licensing and admin toggles should match that intent.
A useful operating sentence for sponsors: We earn the right to widen AI scope the same way we earn autonomy anywhere else—reliability first, then more surface area.
That is calm stewardship. It is also how you avoid painting AI into a corner where turning it off would feel like admitting the intranet failed.
What AI amplifies when the base is weak
When the foundation is incomplete, AI tends to amplify:
Wrong versions — summaries of outdated procedures that still rank in search.
Over-broad context — answers that blend public news with content a role should not see, if permissions were never finished.
Confident tone over accurate ownership — a polished draft that nobody named as accountable.
Shadow process — people trusting a generated answer instead of the controlled page, then emailing the PDF around.
None of that requires blaming the model. The model is doing what the tenant allows. Fix the tenant’s access and IA; then decide where AI may draft, suggest, or summarize under human supervision.
A readiness sequence that respects the calendar
You do not need a multi-year AI program to be ready. You need an honest sequence:
Close the access gaps from Article 1 of this pack—groups, inheritance, existence disclosure.
Stabilize findability from Article 2—hubs, navigation, search, owners.
Name AI use cases that are supervise-first: draft a news summary for an editor, suggest related pages, answer from a known policy library with citation back to the source page.
Define who may turn features on — identity/admin, intranet product owner, and a business risk owner—not a hallway pilot.
Pilot with a narrow audience and an exit criterion: quality, permission incidents, and whether people still open the source of truth.
Keep a reverse switch — documented steps to restrict or disable without a crisis narrative.
If step 1 and 2 are unfinished, skip to them. AI demos will wait.
People still own judgment
AI features in Microsoft 365 can draft, summarize, and retrieve. They do not own policy, risk acceptance, or what “current” means in your organization. Named humans still do.
That stance matches how we talk about AI elsewhere: guardrails and supervision first, autonomy only when earned. The stack here is Microsoft 365. The operating idea is the same. Do not compete with the model on volume of words. Own the map of what AI may touch, who reviews it, and what evidence you keep.
What this is not
This is not a demand to buy more licenses before the intranet works. It is not a Copilot sales script. It is not a claim that McCloy Data is a Power Platform shop or a SharePoint SI factory. It is not résumé language or named individuals. It is public capability language: regulated environments, permission-respecting collaboration, and a bench that has implemented Microsoft 365 intranet foundations so optional AI is a choice—not a cover for unfinished work.
OpenText-on-SAP remains a core landscape for much of our work. Microsoft 365 is often where people open the document. AI readiness belongs on that adjacency only after the base intranet is complete.
A short sponsor checklist
Are access and findability “good enough to defend” without AI in the story?
Which G5 / E5 capabilities you already own are unused for intranet hygiene (identity, compliance, search admin)?
Which AI features are explicitly optional this quarter—and who can reverse them?
For each pilot, who is the human owner of the output?
What would make you pause AI scope—and is that written down?
If those answers are clear, you are closer to AI readiness than any homepage badge.
If you want help sequencing the work
McCloy Data helps organizations finish Microsoft 365 intranet foundations—access, information architecture, findability—and sequence optional AI when the organization is ready. If that is the work in front of you on a tenant you already run, we would welcome a brief conversation.
