Fable 5 for practitioners: hire the model, keep the judgement.

Every solo practitioner runs into the same wall. The work you trained for — the judgement, the relationship, the actual care — sits underneath a pile of work you did not train for: intake summaries, session prep, research you half-remember, follow-up notes written at nine at night. You cannot afford an associate to take that pile. Until now the honest answer was: you just do it, or it does not get done.

Fable 5 changes that arithmetic, but not for the reason the hype suggests. The point is not that the model is smart. Plenty of tools are smart. The point is that it can hold a whole client at once — the intake form, the last three sessions, the labs, the goals — and reason across all of it in one pass. That is not a chatbot. That is the associate you could never justify hiring.

why the big context window is the whole story

A model that can only hold a paragraph forces you to re-explain the client every time. That is exhausting, so people give up and use it for generic tasks. A model that can hold the client’s entire file behaves differently. You brief it once — here is who they are, here is their history, here is what we agreed last time — and every task after that is done in context. The intake summary knows about the labs. The session prep knows about last week. The follow-up email knows what you actually said.

This is the difference between a temp who needs everything spelled out and an associate who has read the file. You are not writing cleverer prompts. You are writing a proper brief once, and then delegating repeatedly against it.

“You would not re-introduce a client to your associate at the start of every task. Stop re-introducing them to the model. Brief it once, then let it work like it already knows the file.”

delegate one: the intake tournament

Do not accept the first summary. Give the model the raw intake and ask for three different ways to frame the client’s presenting picture — different emphases, different hypotheses about what is driving what. Then ask it to argue for the most clinically useful framing and flag what each one might be missing. You are not outsourcing the diagnosis. You are generating a wider set of angles than you would have reached alone at the end of a long day, and then applying your own judgement to choose.

The value is not the winning summary. It is that you considered three before committing to one — the thing good supervision used to provide and solo practice usually cannot.

delegate two: have it interview the file, not you

The second move is to turn the model into a preparation partner. Before a session, brief it on the client and ask it to interview the file: what are the ten questions this history raises that I have not resolved yet? What contradicts what? What did I say I would follow up on and never did? It is not telling you what to do. It is doing the thing a good colleague does over coffee — asking the questions that make you sharper before you walk into the room.

This is also how you protect quality as you scale. The reason care degrades at ten clients is not effort; it is memory. A worker that reads the whole file before every session gives you back the attentiveness of a practice with three clients, at the volume of a practice with thirty.

delegate three: pay it to challenge your plan

Before you finalise a protocol, brief the model to make the strongest honest case against it. Where is the evidence thin? What would a skeptical colleague ask? What is the cheaper or simpler intervention you skipped? This is the red-team move, and it is the one that most quietly raises your standard of care. Not because the model is right — often it is not — but because being forced to answer the objection is how you find the soft spot before your client does.

You keep the plan or you change it. Either way you now understand it well enough to defend it, which is a different thing from having inherited it from a course you took in 2019.

the client-facing rule that keeps you safe

Two lines protect everything above. First, the model drafts and you decide — nothing reaches a client without a human clinician’s sign-off. Second, be honest with clients about where AI helps in your process and where your judgement is doing the work; that transparency is fast becoming a trust advantage, not a liability. Your value was never that you could remember everything. It was that you could be trusted to decide. The model makes the remembering cheap so you can spend more of yourself on the deciding.

“The associate you could never afford now costs almost nothing. What still costs everything — and what clients actually pay you for — is the judgement you put on top of its work.”

what to install this week

Take your next new-client intake. Write one proper brief — who you are, how you practise, what a good summary looks like to you — and delegate the summary, the session prep, and a first-draft follow-up against it. Read each output as a supervisor reading a junior’s work: correct it, keep what is good, bin what is not. Notice how much of your evening you get back, and where your judgement was genuinely needed. That map — delegate here, decide there — is your AI-literate practice, on one page.