Most Copilot deployments fail in one of two ways: they skip the SharePoint permissions audit and start surfacing salary spreadsheets in chat replies in the first week, or they buy 1,000 licenses and stall at 20% weekly active because nobody designed the change management. We do this for a living. The four-phase playbook below is what we ship on every engagement, and the governance pre-flight phase — the long one — is the reason the post-deployment incidents do not happen.
Every page below is comprehensive on its own. Together they cover everything a buyer actually needs to make a Copilot decision — capabilities by surface, industry use cases by vertical, custom agent development, the cross-vendor comparison, and the governance posture that keeps you out of the headlines.
What it does
What Copilot actually does in Outlook, Teams, Word, Excel, PowerPoint, OneNote, Loop, Copilot Chat, Copilot Pages, Copilot Studio, and Wave 3 surfaces (Researcher, Analyst, Agent Builder). For each: prompts that work, time savings reported, where it falls short.
Read the page →Who is using it
What other businesses are doing with Copilot, by vertical. Financial services, legal, healthcare, manufacturing, professional services, sales, marketing, HR, IT, finance, executive admin. Named customers, quantified outcomes, sector-specific cautions.
Read the page →Build custom
The agent factory pattern. The seven canonical agent archetypes (HR onboarding, IT helpdesk, sales SDR, contract review, KB Q&A, BI insights, finance approval). Multi-agent orchestration with the A2A protocol (GA April 2026).
Read the page →GA May 1, 2026
The new agent control plane. Agent Registry · unique Entra identities · Defender runtime protection · Purview data governance · observability dashboard. Manages agents from Microsoft, Salesforce Agentforce, ServiceNow Now Assist, Workday Illuminate, custom builds, and open-source. $15/user/mo standalone, included in M365 E7 Frontier Suite ($99/user/mo).
Read the page →Cross-vendor
Microsoft 365 Copilot vs ChatGPT Enterprise vs Claude Enterprise vs Gemini Workspace. The 6-dimension comparison table. When each wins. The Cowork integration (Anthropic Claude inside Copilot). The 81% of Global 2000 that run multiple model families.
Read the page →Universal
The 6-control AI governance framework we apply to every deployment. NIST AI RMF, ISO/IEC 42001, EU AI Act high-risk classification (August 2, 2026 enforcement). The new SOC 2 question — buyers ask in vendor due diligence.
Read the page →Microsoft 365 Copilot is the flagship, but the Copilot family now includes seven distinct SKUs across productivity, agent orchestration, security, code, and Dynamics. Buying the wrong one — or all of them when you need three — is the most common procurement mistake we unwind in our first call.
$30/user/mo (annual)
What it is: The flagship — Copilot embedded in Word, Excel, PowerPoint, Outlook, Teams, OneNote, Loop, and the standalone Copilot Chat. Grounded in your tenant data via Microsoft Graph.
When it wins: Default for any organization on M365 Business Standard or above. The "knowledge worker productivity" sale.
$5/user/mo or pay-as-you-go
What it is: The free tier renamed and metered. Web-grounded Copilot Chat without tenant grounding. Subset of full Copilot.
When it wins: For users who do not need tenant-data grounding. Often used as the upsell pipeline to full Copilot.
From $200/mo (25,000 messages)
What it is: Low-code platform for building custom AI agents grounded on your data, integrated with 1,500+ connectors. Agent factory pattern.
When it wins: When you need an HR onboarding agent, an IT helpdesk Tier-1, a sales prep assistant, a contract review agent.
$15/user/mo standalone, included in E7
What it is: The new agent-management control plane. Inventory, identity, policy, audit, and cost telemetry for ALL agents in your tenant — Microsoft and third-party.
When it wins: When you have agents from multiple vendors (Salesforce Agentforce, ServiceNow Now Assist, custom Studio agents) and need governance across them.
~$4/SCU/hour, consumption-priced
What it is: AI for SOC analysts. Incident summarization, KQL generation, threat hunting, post-incident write-ups. Native integration with Defender, Sentinel, Entra, Purview.
When it wins: For SOCs running the Microsoft security stack who want to compress Tier-1 triage time and democratize threat hunting.
$39/user/mo
What it is: Coding Copilot with multi-model support (GPT, Claude, Gemini), Workspace task-to-PR agentic workflows, Coding Agent, Code Review, custom knowledge bases.
When it wins: Engineering teams. Pairs with M365 Copilot but is licensed separately through GitHub.
Bundled with Dynamics SKUs
What it is: Sales (account research, pipeline, follow-up drafting), Customer Service (case summary, suggested responses), Field Service (work order intelligence). Microsoft-published outcomes: 20% lower handle time, 30% faster proposals.
When it wins: For Dynamics customers. Often the fastest-payback Copilot surface because the use cases are tightly scoped.
The MSP market has converged on eight named Copilot services (Avanade, Insight, SoftwareOne, CDW, EPC Group all sell some version of these). We ship the same eight. Buyers shopping the market expect to see them named explicitly — so here they are.
2 weeks
Tenant audit, license posture, identity hygiene, SharePoint permissions snapshot, sensitive data discovery, risk-ranked remediation roadmap. Yours to keep whether or not you proceed.
4–8 weeks
Sensitivity labels + auto-labeling, Purview DLP for the Copilot policy location, SharePoint oversharing remediation, Restricted SharePoint Search rollout, retention alignment with PIPEDA / Law 25 / industry obligations.
2–3 weeks
50–200 user cohort selection, persona mapping, role-based training tracks, prompt library curation, success-metric definition (Microsoft's 9 hours/month/user benchmark), GO/NO-GO gate criteria.
Ongoing retainer
Champion network design, prompt library governance, role-based curricula (Sales / Finance / Legal / HR / IT / Marketing), monthly office hours, executive recognition for high-usage teams. Built on the ADKAR change-management spine Microsoft itself endorses.
Monthly retainer
Copilot Dashboard (Viva Insights) telemetry, monthly executive scorecard tied to Forrester benchmarks, cohort analysis, intervention plays for low-adoption groups, quarterly Permission State Report re-run.
Per agent
Copilot Studio agents for HR onboarding, IT helpdesk Tier-1, sales SDR, contract review, knowledge Q&A, finance approval. The agent factory pattern — first agent absorbs the governance scaffolding cost; agents 2–N reuse it. 3–5x ROI uplift over base Copilot.
Per integration
Microsoft Graph connectors (formerly Copilot connectors) into ServiceNow, Salesforce, SAP, ERPs, line-of-business systems. Where the technical depth lives.
4–6 weeks
Copilot Center of Excellence design, AI Acceptable Use Policy, mapping to NIST AI RMF, ISO/IEC 42001, EU AI Act high-risk classifications. The new SOC 2 question — buyers ask about it in vendor due diligence.
Phase 2 is the long one — and the one most MSPs skip. Skipping it is the #1 cause of post-deployment incidents (oversharing, hallucinated content in customer-facing docs, classified-meeting summary surface). We do not skip Phase 2.
Tenant audit, license posture review, identity hygiene check, SharePoint permissions snapshot. We use the Microsoft 365 Copilot Optimization Assessment + SharePoint Advanced Management Permission State Reports to surface the over-permissioning landmines BEFORE Copilot starts surfacing salary spreadsheets in chat replies.
The longest and highest-value phase. Sensitivity-label rollout (or audit-and-extend if labels exist), Purview DLP policy design including the Copilot policy location, oversharing remediation via SAM, Restricted SharePoint Search configuration, retention-policy alignment with PIPEDA / Quebec Law 25 / HIPAA / industry obligations. This is where the audit findings get closed and where the post-Copilot incidents do NOT happen.
50–200 user pilot. Persona-based rollout (Sales gets CRM-anchored prompts; Finance gets Excel patterns; Legal gets summarization workflows). Champion network sourced 3 ways: AI council nomination, manager-selected representative, or top-decile users from the Copilot Dashboard. Prompt library curated and reviewed. Usage measured against the 9 hours/month/user benchmark (Forrester, Microsoft).
Rollout in waves of 500–1,000 users every 2 weeks. Monthly executive scorecard tied to the Forrester ROI benchmarks (132–353% SMB ROI, 116% enterprise ROI with $19.7M NPV per their TEI). Agent factory begins — first agent (typically HR onboarding or IT helpdesk Tier-1) carries the governance scaffolding cost; agents 2–N reuse it. Production-grade rollouts deliver 3–5x the ROI of base Copilot alone.
Below is a preview of the seven verticals we cover in depth on the industries page. Named customer references, quantified outcomes (with vendor caveats), and sector-specific governance considerations.
Earnings transcript analysis, client portfolio prep, regulatory document summarization, FINRA Reg Notice 24-09 record-keeping considerations.
Contract review against playbooks, due diligence summarization, drafting of standard correspondence. Citation work belongs in legal-specific tools (Harvey, Lexis+ AI, CoCounsel) — Copilot for productivity, not for case research.
Administrative workflows. NOT clinical decision-making. DAX Copilot (Nuance) for ambient clinical documentation is a different product family than M365 Copilot but deploys alongside.
Proposal generation from win libraries, research synthesis for engagements, KM ("who at our firm has worked on X"), client meeting prep, time entry narratives.
Operations summaries from sensor data + ops chats, supplier communications, maintenance documentation, safety incident first drafts, multilingual training material across plants.
Pipeline analysis ("which deals slipped this week and why"), follow-up drafting after calls, account research agents, meeting prep one-pagers, proposal generation.
Runbook generation from prior tickets, incident response comms, helpdesk Tier-0 deflection via Copilot Studio, Security Copilot for SOC triage and KQL generation.
Microsoft 365 Copilot is $30 per user per month on an annual commit, layered on top of an eligible M365 plan (Business Standard, Business Premium, E3, or E5). Microsoft 365 Copilot Chat (a subset without tenant grounding) is $5/user/mo or pay-as-you-go. Copilot Studio for custom agents starts at ~$200/mo for 25,000 messages. The new Microsoft 365 E7 Frontier Suite (GA May 1, 2026) bundles M365 E5 + Copilot + Entra Suite + Agent 365 for $99/user/mo — saving roughly $18/user/mo vs. buying components separately. Microsoft 365 E5 is also re-priced to about $60/user/mo starting July 1, 2026.
Copilot honors existing Microsoft 365 permissions verbatim — but it actively surfaces content the user technically had access to but never knew existed. Concentric AI 2026 research found 16% of business-critical data is overshared, with an average of 802,000 files at risk per organization. The fix is a 4–8 week governance pre-flight: SharePoint Advanced Management (SAM) Permission State Report to identify "Anyone" links and broken inheritance, Restricted SharePoint Search to allow-list pre-audited sites for the pilot, sensitivity labels with auto-labeling, and Purview DLP policies for the Copilot policy location to block Confidential content from being grounded. We ship this as a fixed-price engagement before any Copilot pilot user is onboarded.
A standard Kootechnikel Copilot rollout is four phases over 12–22 weeks: Readiness Sprint (2 weeks) for tenant audit and risk assessment; Governance Pre-Flight (4–8 weeks) for SharePoint permissions, sensitivity labels, and DLP; Pilot + Adoption (6–8 weeks) with 50–200 users; then Production rollout in waves of 500–1,000 users every 2 weeks. The Governance Pre-Flight phase is the longest because it solves the oversharing problem before Copilot ever surfaces the content. Skipping it is the #1 cause of post-deployment incidents.
Microsoft and Forrester report 9 hours per month saved per active Copilot user (108 hours/year). Forrester TEI for SMBs found 132–353% three-year ROI; for the composite enterprise, 116% three-year ROI with $19.7M NPV; for sales organizations, 468% high-impact ROI with $47.5M NPV. The risk is not the unit economics — at adoption, $30/user/mo pays back trivially. The risk is adoption itself. Industry data shows the median enterprise hits 30–40% weekly active users out of licensed seats; the bottom quartile sits below 20%. Without a deliberate adoption program (training, prompt library, peer champions, executive modeling), Copilot stalls at ~20–25% weekly active.
Agent 365 (GA May 1, 2026) is Microsoft's new control plane for managing AI agents in your tenant — both Microsoft-built and third-party (Salesforce Agentforce, ServiceNow Now Assist, Workday Illuminate, custom Copilot Studio agents). It provides agent inventory, managed Entra identities for agents, conditional access policies, Purview audit logging of agent actions, and cost telemetry by agent and department. It's $15/user/mo standalone or bundled into M365 E7. You need it when you have agents from multiple vendors operating in your tenant and need governance across them. For an organization with only Microsoft Copilot Studio agents, Studio's native governance is usually enough today.
Copilot Cowork (announced March 9, 2026, expanded into the Frontier program March 30, included in E7 May 1) is Microsoft's integration of Anthropic's Claude models inside the M365 Copilot tenant. In Frontier-program tenants, Microsoft 365 Copilot can route Researcher and Copilot Studio agent reasoning to Claude Sonnet 4.7 or Opus 4.7 instead of (or alongside) OpenAI models. Data stays inside Microsoft's service boundary; Anthropic does not retain or train on tenant data. This is consequential because long-context reasoning and agentic-coding tasks where Claude leads benchmarks can now happen inside the existing Copilot governance perimeter.
81% of Global 2000 organizations now run three or more model families concurrently — "single AI vendor" is not the enterprise reality. Copilot wins when the customer is M365-resident and governance matters; it inherits Entra ID, Purview, DLP, and the entire Microsoft compliance estate. ChatGPT Enterprise wins for general-purpose flexibility and adoption breadth. Claude Enterprise wins for nuanced writing, complex coding, large-document analysis, and structured outputs (which is why Microsoft now embeds it via Cowork). Gemini Workspace wins for Google-native organizations and multimodal RAG. Most clients run Copilot as the safe default with one or two specialist tools layered for specific workflows.
Microsoft 365 baseline data residency is in Canadian data centers (Toronto, Quebec City). Copilot prompt and grounding inference, however, currently leaves Canadian borders — Microsoft has announced in-country Copilot inference for Canada in 2027 (Australia, India, UAE, UK, US ship by end of 2026). Until then, Canadian deployments require documented data-flow disclosures, Quebec Law 25 cross-border transfer notices, and PIPEDA-aligned consent language in the AI Acceptable Use Policy. We map every Canadian Copilot deployment to PIPEDA + Quebec Law 25 obligations as part of Phase 2 governance and have a 2027 in-country roadmap ready for clients who need to plan around it.
Our free 90-minute IT health check includes a Copilot readiness review: tenant license posture, SharePoint permissions snapshot, sensitivity-label coverage, and a phased rollout roadmap with effort estimates. Yours to keep either way.
