The short answer.
Using AI without a plan costs a company five things: seats nobody uses, the same prompt rebuilt in five places, rework when two people’s output differs, know-how that leaves with a person, and decisions nobody made. None of these costs arrives as a single line on a bill, so each one lasts for years. A plan ends all five by naming which work moves onto AI, in which team, and who owns it.
Experimenting was the right way to start. Every person who tried ChatGPT, Claude, Copilot or Gemini on real work taught the company something. The costs below are what experimenting turns into once it is the only thing happening. What happens when a company uses AI without a plan? describes undirected AI use; this piece counts what it costs.
Why the cost of undirected AI is hard to see.
The cost of undirected AI is hard to see because it is spread across every person and every week, in small pieces nobody adds up. The subscription is the only part with a number on it, and the subscription is the smallest part. The rest sits in duplicated effort, corrected work, lost methods and choices left open.
The five costs, one by one.
Each cost of undirected AI has a recognisable sign in a company of under 200 people.
- Seats nobody uses.The company pays for an assistant for everyone. A few people open it every day, most open it now and then, and some have not opened it since the first week. Nobody has told them what AI is for in their role, so the seat waits. Does every employee need an AI licence? covers who needs a seat.
- The same prompt rebuilt in five places.Five people in five teams each work out how to summarise a customer complaint, prepare a weekly report or draft a supplier email. Each one types the background, pastes the records and corrects the first answer. The company pays for the same setup five times.
- Rework when output differs.Two people prepare the same kind of document with different prompts, and the two versions read differently, use different figures or leave out different things. Someone downstream notices, sends it back, and the work is done twice.
- Know-how that leaves with a person.The best method in the company sits in one person’s chat history. When that person changes role, goes on leave or moves on, the method goes with them, and the team starts again from a blank prompt. When the know-how sits with one person covers capturing it before that day.
- Decisions nobody made.Which team goes first, which work moves onto AI, who checks the output, which account company information belongs in: without a plan, each of these is settled by whoever happens to act, or left open. An open decision costs the most of the five, because it keeps the other four in place.
What it feels like from the inside.
Undirected AI use feels exciting at first, then confusing, then quietly tiring. The early weeks bring real finds and real enthusiasm. Then two versions of the same report land on your desk.
People in the company feel it as well. The person with the great method wonders why nobody uses it; the person who never found a use wonders whether they are falling behind. None of this says anything about your people. The experience is chaotic because nobody has yet been given the job of directing it, and that job is easy to hand out once you see it.
The leader’s own position is the hardest. You are asked what the company gets for its AI spend, and the honest answer is “some people use it for some things.” A plan gives you a better answer by next month.
What each cost looks like once there is a plan.
Once there is a plan, each cost has an owner and a next step. The plan is one page per team: the team, its roles, the work that repeats, what moves onto AI first and who owns it.
- Seats.Every seat belongs to a role with named work to run.
- The rebuilt prompt.Work that repeats is written down once, as a workflow the whole team runs in its own ChatGPT, Claude, Copilot or Gemini account. Nobody starts from a blank prompt for that work again.
- Rework.The workflow produces the same shape of output every time, and one named person checks it before it goes out. A workflow prepares; a person decides.
- Lost know-how.The best method in each team becomes that team’s workflow. It stays when the person who found it moves on, and a new hire learns it on day one.
- Open decisions.The plan records each decision with a name and a date: which team first, which work, who owns it, when you look again.
FullyPowered Planner is where that plan lives. You set out your teams, their roles and the work between them, choose what gets built first and name who owns it, and your colleagues add their own teams. When a piece of work is chosen, the workflow is developed from it; FullyPowered Studio captures the method your best person already uses, so the workflow starts from what works in your company.
A cost check to copy.
Copy these questions onto one page and answer them with each team lead. Write “Unknown” where nobody knows, with the name of the person who will find out.
- Seats.Who in this team has an AI seat? For each role, what work is the seat for?
- Rebuilt prompts.Which pieces of work do two or more people prepare with AI, each in their own way?
- Rework.Which AI-prepared output was sent back or corrected last month? Why?
- Know-how.Whose method would the team lose if that person left next week? Where is it written down?
- Open decisions.Which AI question has this team been waiting on someone to answer?
- Owner.Who will own this team’s first workflow?
Every answer becomes a line on the plan. Our AI use is all over the place. What do we do first? sets out the first week.
A 70-person logistics company counts its costs.
Take a 70-person logistics company with four teams: dispatch, customer service, billing and the warehouse. The operations director ran the cost check with each lead.
Illustrative example.
- Dispatch.Every planner has a seat. Two use it daily to draft delay notices to customers, each with their own prompt, and the notices read differently. The cost: rebuilt prompts and rework.
- Customer service.One agent built a method for summarising damage claims from photos and notes. Nobody else knows it, and the agent is moving to the sales team. The cost: know-how about to leave.
- Billing.Seats were bought for everyone, and nobody uses them. Nobody has decided whether AI may prepare invoice queries at all. The cost: seats nobody uses, held there by a decision nobody made.
- The warehouse.No seats and no use. The shift handover notes repeat every day and start from the warehouse system. The cost: nothing spent, and nothing gained.
The operations director put all four teams on one plan. Customer service went first: the damage-claim method became the team’s workflow before the agent moved, with the customer service lead checking every summary. Dispatch went second, with one shared workflow for delay notices. Billing’s open decision got a name and a date.
What is different in a company of under 200 people.
In a company of under 200 people, the costs of undirected AI land on fewer people, so each one hurts more. A large company has spare hands to absorb rework and teams to rebuild lost methods. You have one person who knows the method and one lead who checks the work.
- One departure empties a team’s know-how.When the one person with the method leaves a team of four, the team loses it entirely.
- A plan fits on one page.No committee, no assessment. One owner, one page, a decision for each team.
Name the five costs in your own company, then put them on one plan. How organizations plan their use of AI covers the method, and FullyPowered Planner is where you set out your teams and choose what to build first.