What does a chat assistant subscription provide?
A chat assistant subscription provides each person with a capable general-purpose tool: ask it to draft, summarise, analyse, or explain, and it does, well, for whatever the person is working on at that moment. It is enormously useful for the people using it, and most organizations should provide one.
Its structure is the constraint. It does nothing until a person prompts it, it holds no responsibility for what happens with its output, and the context it has is whatever the person gave it in that conversation.
What does a Digital FTE provide?
A Digital FTE provides a worker for a function. It starts each day with a morning brief of what it plans to do and what is waiting on the manager, works inside your systems through scoped connections, proposes every action with evidence through an approval gate, and is reviewed against explicit targets in its Performance tab.
Its context is a Brand Brain memory of your business, with each item's provenance visible and confirmed by you, rather than whatever fits in a conversation. And its accountability is structural: a named person approves each action, and the audit trail records it.
Why is 'we already have ChatGPT' not the same?
It is not the same because having a capable tool does not give a function an owner. A support queue does not get resolved because the team has an assistant; someone still has to open each ticket, prompt the assistant, check the answer, and send it. The person is the worker and the bottleneck, and nothing happens when they are away.
A Customer Support Digital FTE resolves the queue on its own cadence, cites its sources, escalates the fixed list, and drafts replies for approval. The function has an owner, and the person becomes its manager rather than its operator.
- Initiative: prompted versus self-starting within a remit
- Systems: pasted context versus scoped connections to your tools
- Memory: conversation versus confirmed Brand Brain with provenance
- Accountability: the user's versus a named approver in the audit trail
- Measurement: none versus targets and a performance view
How do governance and cost differ?
Governance differs because an assistant's output goes wherever the person sends it, with no gate, no fact gate, and no trail beyond the conversation history. A Digital FTE's output passes the fact gate for anything published and the approval gate for anything that acts, and every decision is logged with who and when.
Cost differs in structure. An assistant subscription is per seat, attached to people. A Digital FTE is metered work from one organization wallet, attached to the function, with a hard stop when the wallet is empty.
What is the right combination?
The right combination is both. Give people a chat assistant for their own work, because it makes them faster. Hire Digital FTEs for the functions that need an owner, because an assistant cannot be one. The two do not compete; a team with a good assistant is usually better at managing a Digital FTE, because they already understand what a model can and cannot do.
The test for any function is whether it has a person who should keep doing it. If yes, that person gets an assistant. If the function has no proper owner, or its owner is spending their judgement on repetitive work, it gets a Digital FTE.
What does a team gain from having both?
A team with both gets two different things and, less obviously, gets better at managing the second because of the first. The chat assistant makes each person faster at their own writing, analysis, and correspondence, and teaches them, through daily use, what a model does well and where it invents. That experience is exactly what a manager needs when reviewing a Digital FTE's proposals: they know to check the evidence rather than the fluency, and they recognise a plausible-but-unsourced claim on sight.
The Digital FTE gives the function an owner that works on its own cadence, proposes with evidence through a gate, holds a confirmed memory, and is reviewed against targets, which no amount of prompting can produce from an assistant. The two costs sit on different lines, per seat and metered work, and do not compete. The test for any piece of work remains the same: if a person should keep doing it, give them the assistant; if the function has no owner or its owner is spending judgement on repetitive work, hire the Digital FTE and manage it.