The consultant you can clone

Why this matters to you right now

Jamin Ball, in his "Digital Twins" analysis published on Clouded Judgement in March 2026, identifies six flavors of digital twin that have emerged among funds investing in enterprise AI. For a solo consultant or a boutique practice, three flavors are directly operational today: workflow capture (the processes that live in your head), institutional memory (what you know about a client after two years of engagement), and knowledge multiplication (the "one to many" that turns your expertise into leverage).

Most consultants picture the digital twin as a chatbot their clients could query. That's the wrong direction and it is what makes them wait. The twin that actually changes a practice is internal. The client never talks to it. You do. It handles the work that stalls on you: prospect triage, first-draft offers, brief extraction from discovery transcripts, review of your own documents. It is half of your day turning into a productive asset instead of a bottleneck.

Edition #14 on AI for senior consultants argued that AI amplifies an existing discipline. This edition shows the finished form of that amplification.

If your practice only works when you are at your desk, you don't have a practice. You have a bottleneck with a business card.

Pricing: the fixed fee that the twin makes defensible

The action: Identify your low-variance deliverables (standardized diagnostics, recurring audits, quarterly mini-reports) and re-price them as fixed fees, on the basis that your twin produces 80% of the first draft and you finalize with your judgment. The fixed fee doesn't go down. It goes up, because delivery lead time compresses and net margin increases.

Why it works: Jamin Ball calls this category "knowledge multiplication," the one-to-many mechanic. Once your judgment is codified in a twin, the marginal cost of the tenth deliverable approaches the marginal cost of the first. The client still buys your name and your final sign-off. They no longer pay you for the six hours of manual first draft that would otherwise have consumed your Monday. The fee reflects value, not effort.

The trap: Pricing on cost ("my twin runs for a few dollars, I can lower the fee"). Classic mistake. The price of a deliverable was never the cost of its production, it is the value perceived by the client. A deliverable produced in two hours with your twin bills exactly the same as a deliverable produced in eight hours without, because it is the same strategic value for the client. The logic is exactly the one in edition #4 on delivering less, charging more: billing follows the result, not the input.

This week: Identify one single deliverable you produce recurrently (at least three times in H1). Write the precise list of inputs required to produce it: client information, contextual data, constraints. That list becomes the specification of your first twin workflow.

Sales and development: the triage that costs nothing

The action: Build a triage agent for your inbound conversations. You feed it the five questions you ask every prospect on the first call, plus your qualification criteria (minimum engagement size, sector, type of problem). The agent receives the contact form or the discovery call transcript and produces a qualification dossier with a recommendation: pursue, decline, redirect.

Why now: In a typical consulting practice, between 25% and 40% of inbound conversations lead nowhere. They are hours that evaporate in politeness. Automated triage doesn't eliminate those hours, it relocates them: the twin sorts in fifteen minutes, and you only give your time to already pre-qualified conversations. The result isn't more engagements. It is fewer hours wasted before every signed engagement.

I tested exactly this last month on the inbound requests through my site. The twin classified each request into three buckets: respond directly with a precision question, recommend a discovery conversation, or recommend declining or redirecting. Across 23 requests processed in May, the twin recommended declining 6 times. Six 45-minute conversations I didn't have. Three hours recovered without losing any of the real pipeline.

The trap: The twin that is too permissive (says yes to everything) or too conservative (says no to everything). The initial calibration requires showing it 10 to 15 past examples with your real verdict. Without that calibration, the twin produces generic recommendations that don't reflect your personal threshold.

This week: Write the five exact questions you ask every prospect on the first call, plus the sentence that describes your ideal-engagement criterion. Submit that to Claude along with ten recent prospect examples (known outcome: signed, declined, lost). You have version zero of your triage agent.

Collaboration networks: the twin shared between complementary peers

The action: Identify a peer whose practice is complementary to yours (a strategy consultant working with an execution consultant, for example). Together, build a twin that serves as an interface: your twin handles intake and preliminary diagnosis, theirs takes over for deliverable design. The client sees a unified experience. Both of you keep your independent practices.

The mechanism: The Block and Sequoia manifesto From Hierarchy to Intelligence argues that traditional organizations are first and foremost information-routing systems bounded by the human span of control. AI can replace that routing function inside an organization, and between organizations that choose to collaborate. For two solo consultants who will never merge into a single firm, the shared twin is the cheapest form of merger. You keep your brand, your pricing, your clientele. You only share the intake agent and the coordination layer.

The trap: Building two isolated twins that don't talk to each other. That is the technological version of the ball you bounce back and forth by email. The benefit disappears if each peer re-keys the information into their own system. The logic is exactly the one in edition #25 on the client who brings three: what creates leverage is explicit orchestration, not the intent to collaborate.

This week: Identify the candidate peer. Sketch on one page what your shared twin would do in the concrete case of one single typical prospect: who takes the lead when, who validates what, where the information lives. One page is enough to decide whether the conversation deserves to go further.

Value creation: codify so you can clone

The action: The twin doesn't fabricate spontaneously. It operates on codified inputs. If one of your work phases isn't codified (method, deliverables, acceptance criteria), the twin will produce a convincing and false first draft. Before writing a single prompt, codify the phase you want to clone. One page is enough for a typical phase.

Why it changes everything: This is the conceptual heart of the edition. Edition #6 on methodology as your product argued for codification as building intellectual property. At the time, the argument was commercial. Today, it becomes operational: codification is the technical precondition for a twin to exist. Tiago Forte, in Building a Second Brain, states the idea in the positive: everything you capture outside the brain becomes an infrastructure you can activate later. For a consultant, that infrastructure becomes the soul of the twin. No capture, no twin. With capture, the twin amplifies what you have already understood.

I personally postponed codifying one of my diagnostic phases for three years, telling myself it was too unique to deserve a template. The day I tried to hand it to an agent, I had to reconstitute in two weeks what I could have written in one afternoon at the time. Postponed codification is paid at full price the moment you reach for leverage.

The test: How many phases of your work can you describe to a junior in less than 30 minutes, without them walking out with substantive questions? That number is the number of phases a twin can start executing for you tomorrow. All the others are candidates to codify this week, before the twin. The logic extends edition #26 on the calendar that eats your plan: the protected weekly block you may just have defended only makes sense if you use it to do the codification that makes the twin possible.

This week: Pick ONE phase. Write on one page: its inputs, its steps, its quality criteria, its output artefact. That page becomes the system prompt of your first twin workflow next week.

AI: build your first twin this week

The cost of a consultant's twin in 2026 isn't technical anymore. It is intellectual. The tools exist, are mature, and cost less per month than a client lunch. The only cost is the discipline of codification.

The action: Build your first minimal twin this week. Five components are enough for version zero, accessible today with no infrastructure.

1. 1.The triage agent (the use case from section 3). Receives an inbound message, classifies into three categories, proposes a first-level response. This component alone typically saves two to four hours a week.

1. 1.The brief extractor. Give it a discovery call transcript (Otter, Fireflies, Granola). It produces a structured brief: client objectives, constraints, buying signals, open questions, proposed next steps. This component turns 90 minutes of manual note-taking into five minutes of review.

1. 1.The first-draft offer writer. Give it the brief produced by component 2, plus your offer template, plus your last five signed offers. It produces a first draft of 70% to 80% quality that you finalize. The codification of your method (section 5) is what makes the difference between a generic draft and a draft that already sounds like you.

1. 1.The document reviewer agent. Submit your draft offer. It checks for the three-marker Definition of Done (see edition #24 on the offer that bleeds you), flags grey zones, proposes rewrites. You validate or not.

1. 1.The weekly strategic review agent. The component introduced in edition #26. Reads your calendar, your CRM, your issued invoices, and flags drift relative to your OKRs. Serves as a silent auditor that keeps the mid-year pivot from being a surprise.

Where most consultants get it wrong: Believing the twin has to be impressive before it is useful. False. The version zero of each component above is imperfect and already delivers saved hours from the first week. The twin refines itself continuously through use, not through preparation. Reid Hoffman and Greg Beato, in Superagency, state the principle: AI augments human agency when it is put into service early and corrected through use, not when it is planned to perfection on a whiteboard.

The warning: The twin amplifies an existing discipline. If your method isn't codified (section 5), the twin will produce convincing and false outputs. The cost of a bad offer draft generated quickly and sent to a client is higher than the cost of having no twin. Codify first. Build second.

This week: Pick one single component out of the five. Not all five. Give yourself four hours of codification (section 5) plus two hours of build. Version zero is functional and already useful next Monday.