The Three-Phase Journey to Becoming a Frontier Firm

You understand what a Frontier Firm is. You know the Microsoft technology stack that powers them. Now comes the crucial question: How do you actually transform?

The good news: successful organisations follow a predictable pattern. Microsoft’s research reveals three distinct phases that organisations progress through, each with specific objectives, timelines, and technologies.

The better news: you don’t need to have everything figured out on day one. Different parts of your organisation can progress at different speeds.

Let’s break down each phase with practical timelines, key actions, and real-world examples.

Phase 1: AI as Assistant (Months 1-4)

The Objective

Make every employee more productive by automating repetitive, time-consuming tasks. This phase is about proving value quickly and building organisational confidence in AI.

Key Actions

Deploy Microsoft 365 Copilot to pilot groups (50-100 users initially). Don’t try to boil the ocean. Select diverse users across functions who will be honest about what works and what doesn’t.

Identify high-value use cases: Email management, meeting summarisation, document creation, and data analysis. Focus on tasks that consume significant time but don’t require deep expertise.

Establish governance framework: Usage policies, data security controls, compliance requirements. Yes, this feels like overhead when you’re excited to get started. But trust me, you don’t want to roll out broadly and then discover compliance issues.

Create measurement dashboard: Track time savings, output quality, and user satisfaction. You need data to justify expansion and secure additional investment.

Launch training programs and champion networks: Technology doesn’t adopt itself. Identify enthusiastic early adopters who can help their colleagues.

Technology Focus

Microsoft 365 Copilot across Word, Excel, PowerPoint, Teams, and Outlook. Please keep it simple. Master the foundation before adding complexity.

Expected Outcomes

Measurable time savings on routine tasks are typically 30 minutes to an hour per employee per day. Improved document quality with fewer revision cycles. Growing user confidence and enthusiasm.

Most importantly: proof points. You need concrete examples of value to secure broader rollout.

What Success Looks Like

At the end of Phase 1, you should have:

• 80%+ of pilot users actively using Copilot weekly
• Documented time savings with dollar values attached
• A library of successful use case examples
• Executive buy-in for broader deployment
• A clear roadmap for Phase 2

Real-World Example

Adecco achieved up to 63% productivity increase during their Microsoft 365 Copilot rollout. They focused initially on high-volume activities like email management and document creation, proving value before expanding to more complex use cases.

Phase 2: Agents as Team Members (Months 5-10)

The Objective

Create AI agents that function as digital colleagues, handling specific tasks and subtasks with human direction. This is where things get interesting and where most organisations currently stand.

Key Actions

Build department-specific agents in Copilot Studio. Start with processes that are high-volume, rule-based, and well-documented. IT help desk. HR onboarding. Sales qualification. Customer support triage.

Integrate agents with knowledge sources: SharePoint, OneDrive, and databases. Your agents need to understand your organisation’s specific context, not just general knowledge.

Add action capabilities through connectors, flows, and APIs. Agents that only answer questions have limited value. Agents that can actually execute tasks create tickets, update records, trigger approvals, and deliver transformational impact.

Deploy agents to Teams for seamless collaboration. Agents shouldn’t live in separate applications. They should be where your employees already work.

Establish agent monitoring and quality assurance processes. How do you know if agents are performing correctly? How do you catch errors before they impact customers? Build these systems early.

Train employees in “agent boss” skills: Delegation, prompt refinement, output review. This is a new skill set for most employees. Provide structured training and ongoing support.

Technology Focus

Microsoft Copilot Studio, Power Platform, and Azure AI services. You’re now building custom solutions tailored to your business.

Expected Outcomes

Faster workflow cycles typically 20-30% improvement. Measurable back-office efficiency improvements. Reduced time-to-resolution for customer inquiries.

Employees begin thinking differently about their work. Instead of “How do I complete this task?” they ask, “How do I delegate this to an agent?”

Real-World Examples

ClearBank reduced payment recovery processing time by 80% using Azure AI Document Intelligence integrated with custom agents. They started with a pilot handling 100 cases per week, proved the model, then scaled to thousands.

ClearTax enabled 75% of users to file taxes for the first time, unlocking INR 300 million in refunds through Azure OpenAI Service-powered agents. They built agents that guided users through complex tax regulations in natural language.

Datasite reduced redaction times by up to 80% for M&A lawyers using Azure AI. Their agents handle the tedious work of identifying and redacting sensitive information, while lawyers focus on strategic document review.

What Success Looks Like

At the end of Phase 2, you should have:

• 5-10 production agents handling real business processes
• Clear documentation of ROI for each agent
• Employees actively proposing new agent use cases
• Established governance and monitoring frameworks
• Technical capability to build new agents in weeks, not months

Phase 3: Human-Led, Agent-Operated (Months 11-18+)

The Objective

Enable AI agents to autonomously manage entire business processes from end to end, with humans providing strategic oversight. This is the actual Frontier Firm state.

Key Actions

Deploy autonomous agents across Dynamics 365 for sales, service, finance, and operations. These aren’t simple chatbots, they’re sophisticated agents that handle complex workflows with minimal human intervention.

Implement MCP servers to expose business capabilities to agents. Make it easy for agents to interact with your core business systems using standardised protocols.

Establish flexible, outcome-driven team structures. Traditional org charts don’t make sense when half your “team” consists of AI agents. Redesign around outcomes, not headcount.

Define human-agent ratios for different roles and departments. How many agents should each sales rep manage? How many customer service agents per supervisor? These ratios become as crucial as traditional span-of-control metrics.

Create continuous improvement loops with evaluation dashboards. Agents should get better over time. Build systems that identify underperforming agents and optimise their behaviour.

Develop advanced agent orchestration across multiple systems. The most valuable agents don’t live in isolation; they coordinate with other agents to accomplish complex objectives.

Technology Focus

Dynamics 365 with agentic capabilities, Azure AI Foundry Agent Service, and MCP integration. You’re now operating at the frontier of what’s possible.

Expected Outcomes

Significant acceleration of business processes by 30-50% in many cases. Reduced human error. 24/7 operation capability without proportional headcount increases.

Perhaps most importantly, your organisation begins innovating in ways that weren’t previously possible. When intelligence is abundant and cheap, what becomes possible?

Real-World Examples

Husqvarna Group reduced technician resolution times by over 60% with the AI factory companion integrated into their operations. Agents diagnose equipment issues, recommend solutions, and even order parts all autonomously.

NTT DATA achieved 65% automation in IT service desks and up to 100% automation in specific order workflows. Their agents handle everything from password resets to complex provisioning tasks.

ABB achieved 35% cost savings in operations and a 30% production efficiency boost. They’ve deployed agents that monitor production lines, predict maintenance needs, and optimise scheduling, running 24/7 without human supervision.

What Success Looks Like

At the end of Phase 3, you should have:

• Multiple business processes running with 50%+ agent automation
• Human employees focused on strategic work, not operational tasks
• Clear competitive advantages in speed, cost, or quality
• Organisational culture that embraces human-agent collaboration
• Recognition as a leader in your industry’s AI transformation

Common Questions About the Journey

Q: Do we need to complete Phase 1 before starting Phase 2?

Not entirely. Your IT department might be ready for Phase 2 agents while your finance team is still in Phase 1 with Copilot. Progress at different speeds based on readiness and use case maturity.

Q: How long does the whole journey take?

Most organisations spend 12-18 months moving from Phase 1 to Phase 3. But remember: this isn’t a destination. Frontier Firms continuously evolve their agent capabilities.

Q: What’s the most prominent reason organisations stall?

Lack of clear success metrics. If you can’t demonstrate ROI from Phase 1, you won’t get budget for Phase 2. Measure everything.

Q: How much does this cost?

Variable, but think of it as reallocating existing labour costs rather than pure new investment. Yes, you’ll pay for Copilot licenses, Copilot Studio capacity, and Azure resources. But you’re typically saving multiples of that in efficiency gains.

The Critical Success Factor: Momentum

Here’s what separates successful transformations from stalled pilots: momentum.

Organisations that succeed move quickly from proof-of-concept to production. They celebrate early wins publicly. They invest in training and change management. They make AI transformation a strategic priority, not an IT side project.

Organisations that struggle get stuck in endless pilots, waiting for perfection before scaling. They underinvest in change management. They treat AI as a technology initiative rather than a business transformation.

Which organisation will yours be?

What Comes Next

Understanding the three-phase journey provides the roadmap. But execution requires more than a map; it requires concrete strategies for overcoming challenges, building capabilities, and measuring success.

In our next post, we’ll explore five essential implementation strategies that separate successful Frontier Firm transformations from failed pilots.

About This Series: This is the third of five articles exploring how organisations can transform into Frontier Firms using Microsoft technology. In the next instalment, we’ll detail essential implementation strategies and how to overcome common challenges.

Original Post https://daveburrell.co.uk/the-three-phase-journey-to-becoming-a-frontier-firm/