283,000 Consultants Just Got AI Agents — Independent Practices Have 6 Months

KPMG is embedding Claude into Digital Gateway, its Azure-based platform, across all 138 countries and 276,000 employees. Tax agents that used to require weeks now build in minutes. Managed Agents handle repeatable work. KPMG named Anthropic preferred PE partner for portfolio deployment. This is not a pilot. It is doctrine.

Bain named Anthropic Global Premier partner and embedded Claude across 7,000+ employees within weeks of pilot launch. Two-thirds adopted the Excel add-in. Thirty to fifty percent productivity uplift on complex code. Fifteen hundred AI and data experts now run agent architecture across the firm.

Total: 276K KPMG plus 7K Bain equals 283,000 consultants working behind AI agents.

Your independent practice still runs email. Zoom. Google Docs. This is not a technology gap. This is a speed gap. When a Big Four team builds a tax agent in minutes that used to take weeks, that is not innovation. That is a price compression event.

How Big Four Speed Kills Independents

Speed reduces cost. Cost reduces fees. Reduced fees kill margins for anyone competing on commoditized work.

Before AI agents, a tax strategy project took months. Client hired a Big Four team. The team billed $500K. Half the hours were junior staff learning the codebase, pulling data, formatting reports, testing edge cases. Those are the hours that AI agents now execute in the engine room—invisible work that used to drive billable margin.

Now KPMG builds the tax agent once. The agent handles data extraction, analysis, edge-case testing, report generation. The client gets the deliverable in two weeks instead of three months. The Big Four team writes the agent code. The agent does the repetitive work. The client pays less. The Big Four keeps more margin because labor cost collapsed.

Where does that compress fees? On your clients' desks. They see KPMG delivering the same work in a quarter of the time. They ask you why you take three months. You tell them it takes that long. They tell you they will call KPMG.

KPMG's Blaze modernizes IT through Claude Code, embedding agent-assisted development into portfolio companies. That is not consulting. That is doctrine transfer. KPMG is building a system where their PE portfolio operates faster than the market baseline. That advantage compounds across every portfolio company. That is competitive moat construction: not legal moat. Operational moat.

The 90-Day Bottleneck Audit

You have six months before this hits your target market. Do a bottleneck audit now. Find the work that kills your billable hours and eats margin.

Step one: Catalog your current production process for your flagship service. If you are a fractional CFO, map the budget cycle. If you are a tax consultant, map the audit prep. If you are a strategy consultant, map the competitive analysis and presentation build. Write down every task. Estimate hours.

Step two: Mark which tasks are learning work, which are mechanical, and which are judgment.

Learning work is your job. You hire for it. It cannot be automated. But you can augment it.

Mechanical work is where AI agents will eat you first. Data extraction, formatting, validation, testing, reporting: these are the hours Big Four eliminated. Your clients will not pay you for mechanical hours once they know it takes an agent 30 minutes.

Judgment work is what you sell. The strategy recommendation. The financial structure. The risk callout. The client pays for this. But if you are the bottleneck on all three tiers, you cannot scale without hiring, and hiring kills margin.

Step three: Build a test agent for one mechanical task. Not a chatbot. An agent that takes structured input, does the work, and delivers the output. Tax agent that pulls K-1s and schedules from documents and generates a summary. Valuation model that pulls financials and comps and outputs a range. Budget model that takes historical data and outputs a scenario matrix.

Do not spend three months on this. Spend two weeks. Use Claude and ask it to help you architect the agent. Describe the input and the output. Let the model walk you through the steps. Build a prototype.

Step four: Measure how much time this agent saves you per client. Multiply by your client count. That is your runway before margin compression hits.

Speed Is Competence in Disguise

The Big Four just redefined what competence means. It no longer means expertise plus hours. It means expertise plus systems. A KPMG tax team working behind an agent is faster, cheaper, and more accurate than a solo tax consultant without one. That is not opinion. That is the engine room floor.

Your credential is not your moat anymore. Your process is. Your system is. Your ability to automate the mechanical work and focus on judgment is.

That is not advice. That is the market clearing.

How Big Four Scales the Moat

KPMG is not just embedding Claude. They are building AI-native operations architecture. When a portfolio company needs to modernize its tax process, KPMG does not send consultants. They deploy the tax agent. The agent handles the mechanical work. A fractional oversight team handles the judgment work. Cost per project drops. Revenue per project stays flat. Margin expands. Portfolio company gets better service cheaper.

That is not consulting. That is ownership-class operations. KPMG is building the system that an owner-operator would build if they had the time and capital. They have both. You do not.

So you have two paths. Build your own AI operating system and compete on speed and margin. Or stay on the email-and-Zoom model and watch fees compress.

You have six months before the Big Four moves down-market and your clients start hearing about this moat.

Doctrine Connection: Competence Beats Credentials

Your MBA does not beat KPMG's tax agent. Your CPA does not beat Bain's code agent. Your certification does not beat a system that learns from ten thousand prior engagements.

The market is not asking for better credentials anymore. It is asking for faster delivery. Cheaper delivery. Fewer hand-offs. Better accuracy. That is what systems deliver.

Credentials open the door. Competence: measured in speed, cost, and accuracy: keeps the door open. Systems are competence at scale. Build the system and the competence compounds. Do not build it and watch the credentials stop mattering.

FAQ

Q: Can I compete with KPMG if I build an AI agent? No. But you can compete with the 95% of firms that have not built one yet. That is a six-month window. Use it. KPMG has capital, brand, and distribution. You have speed and agility. Build the agent and you move faster than they do on your vertical. That is the moat.

Q: What if my clients do not care about speed? Then you are in a different market. They care about outcome. Build an agent that improves outcome. Better financial recommendations. Better tax strategy. Better risk assessment. Speed is the engine. Outcome is the hull. You need both.

Q: How much does it cost to build an AI agent for my service? Proto to production: two to six weeks depending on complexity. Labor cost: one developer or you building it yourself. Tool cost: Claude API is $3-20 per task depending on complexity. Most solo consultants can build their first agent for under $5K in labor and tool cost. Most firms with dev resource can do it faster and cheaper.

Q: What happens after I build the agent? You have three options. One: keep it internal and raise your margin. Two: resell it to other consultants on your vertical and build a SaaS margin. Three: give it to your clients and shift to outcome-based fees where you share the efficiency. KPMG chose three. You choose based on your capital structure and ambition.

Jeff Barnes has no personal position in any company named in this article. DEMG provides marketing systems, not investment advice.

The Speed Compression Timeline

The numbers tell the story. KPMG says building a tax agent that used to take weeks now takes minutes inside Digital Gateway. That is not a feature announcement. That is a price signal.

When a Big Four team produces a deliverable in minutes that an independent practice produces in days, the client sees two facts. The deliverable quality is comparable. The speed is not. The independent practice must either match the speed or justify the gap with something the Big Four cannot deliver.

According to Statista, the global management consulting market reached approximately $330 billion in 2025. The Big Four collectively control the largest share. Independent practices survive on relationships, specialization, and speed of engagement. AI agents compress speed advantages.

This is not an argument against independent consulting. It is an argument for building your own AI operating system before the competition builds it for you.

The 90-Day Bottleneck Audit for Independent Consultants

Apply the framework. Four steps. Ninety days.

Week 1 through 2: Map every repeating deliverable. List every document, analysis, report, and presentation your firm produces more than once per quarter. These are your automation candidates.

Week 3 through 4: Identify the time bottleneck in each. For each repeating deliverable, where does the clock stop? Data gathering? First draft? Client review cycles? Internal review? The bottleneck is the automation target.

Month 2: Build one AI agent for the largest bottleneck. Not a chatbot. An agent that takes structured input, produces a first draft of the deliverable, and routes it for human review. Anthropic Claude documentation provides the building blocks. The agent does not replace judgment. It replaces the blank page.

Month 3: Measure and iterate. Track time to first draft before and after. Track revision cycles. Track client satisfaction. If the agent saves 40% of delivery time on your highest-volume deliverable, scale to the next one. If it does not, adjust the agent and retest.

The goal is not to compete with 276,000 people. The goal is to match the speed while keeping the relationships and specialization that clients chose you for in the first place.