The problem.
What companies are dealing with.
- How are companies investing in AI, and which part is nobody managing?
AI spending in a company goes two places: a few new functions that AI runs, and an assistant such as ChatGPT, Claude, Copilot or Gemini for nearly every employee. The new functions usually get a budget, a project plan and an owner. The assistants reach most of the people and, in most companies under 200, take most of the AI spend, yet they were switched on with no plan and no owner, and most people use them only as a chat box.
Full article: The two kinds of AI investment, and the one nobody is managing.
- Everyone at our company already uses ChatGPT. Doesn’t that mean we know AI?
Being able to type a question into ChatGPT, Claude, Copilot or Gemini and get a fluent answer is not the same as knowing AI. Chatting is only the surface of those assistants. The same account a company already pays for can hold a team’s instructions, read its files, work from its records and run a piece of work the same way every time, and most people never reach that part, so a company where everyone chats can miss it for years.
Full article: Everyone can chat. That is not the same as knowing AI.
- Why do some people get so much more out of AI at work than others?
The gap in AI at work looks like the gap between basic and advanced Excel. Most people use ChatGPT, Claude, Copilot or Gemini the way a beginner uses a spreadsheet: type something in and read what comes back. The people who get the most from an AI assistant have built reusable workflows that prepare whole pieces of work the same way every time, so most companies are paying for the advanced tool and using the beginner’s version of it.
Full article: Basic Excel and advanced Excel: the same gap exists with AI.
- What does advanced use of ChatGPT, Claude, Copilot or Gemini actually look like?
Advanced use of an AI assistant such as ChatGPT, Claude, Copilot or Gemini is everything past the chat box. It means workflows saved once and run again, projects that hold standing instructions, connections that let the assistant read from business systems, and runs that happen on a schedule. Nearly everyone at work can type a prompt, but most people have never opened any of those four kinds of feature, so most companies use a powerful tool as a calculator.
Full article: What “advanced” use of ChatGPT, Claude, Copilot or Gemini actually is.
- Why isn’t AI saving us time?
AI is not saving most companies time because people use it on small parts of the job, while the time in a business sits in work that repeats every week. Four things keep everyday use from adding up: it is used for tasks rather than whole pieces of work, everyone starts from a blank prompt, the checking was never designed, and the time saved has no destination. The fix is to point the AI already paid for at named, repeating work with an owner.
Full article: Why isn’t AI saving us time?
What it costs to leave it.
What happens when it is left as it is.
- What happens when a company uses AI without a plan?
When a company uses AI without a plan, its AI use is undirected: each person works AI out alone, and most end up using ChatGPT, Claude, Copilot or Gemini only to ask questions, draft and summarize. The larger gain, handing whole pieces of repeat work to AI team by team and role by role, never arrives, because nobody decided it should. The company keeps paying for seats and cannot say what they bought.
Full article: What happens when a company uses AI without a plan?
- How can I tell if our company’s AI use is undirected?
The clearest sign of undirected AI use in a company is that nobody can say which work AI is meant for: the same task is done a different way by each person, and a good method stays with whoever found it. To check, spend one week asking and looking for ten plain signs, then count how many appear. Three or more signs mean the company’s AI use needs a plan; the count is no judgement on the people doing the work.
Full article: How do I know if our AI use is undirected?
- Everyone at our company uses AI differently. Is that a problem?
Yes, and the trouble shows up between teams. When everyone uses AI their own way, handoffs stop matching because each team changed its output alone, customers get two answers to one question because two teams drew on different sources, and checks get skipped because each person assumed someone else checked. The fix is a plan for the work between teams: for every handoff, one agreed format, one source for the facts and one named person who checks.
Full article: What goes wrong when everyone uses AI their own way?
- What does it cost a company to use AI without a plan?
Using AI without a plan costs a company in five ways: paid seats nobody uses, the same prompt built again in several places, rework when two people’s output does not agree, know-how that walks out when a person leaves, and decisions nobody ever made. None of these appears as one line on a bill, which is why they can last for years. A plan ends all five by naming which work moves onto AI, in which team, and who owns it.
Full article: What does it cost a company to use AI without a plan?
- Who is responsible when AI gets something wrong at work?
In a company without an AI plan, nobody is responsible when AI gets something wrong, because nobody was ever told the work was theirs to stand behind. In a company with a plan, every workflow carries two names before it runs: an owner, who keeps the workflow right and fixes it when it misses, and a checker, who reviews each piece of work and decides before it leaves the team. AI prepares the work; a person decides.
Full article: Who answers when AI gets something wrong?
What a plan is, and who owns it.
What a plan is, what goes in it and who owns it.
- What is an AI plan for a company, and what goes in it?
A company’s AI plan is one page that lists its teams, the roles in each, the work each team repeats and the first workflows each team should have, with a named owner and a named decision-maker for every workflow. It settles, team by team, which work AI should prepare, who owns each piece and what gets built first. The plan is the first of three steps: plan it, develop the workflows, deploy them. FullyPowered AI Planner is where that plan lives.
Full article: How organizations plan their use of AI.
- What is the difference between an AI strategy, an AI policy and an AI plan?
The three documents answer three different questions. An AI strategy answers why a company uses AI and where it matters most. An AI policy says what is allowed: which assistant, what information may go into it, and that a person decides. An AI plan says which team and which role gets which workflow, in what order, and who owns each by name. A company under 200 people needs a strategy of a few sentences, a two-page policy and a plan it keeps current, because only the plan tells a team lead what to do next week.
Full article: AI strategy, AI policy, AI plan: what is the difference?
- Who should own AI in a small or mid-sized company?
In a company of 10 to 200 people, AI is owned by one named person who already runs the work: the head of operations, the chief of staff or the owner, rather than a committee, a new head of AI or IT alone. That person owns the company’s AI plan, keeps it current, settles the order of work between teams and runs a monthly review. IT handles accounts, access and the approved assistant, and each workflow has its own owner in the team that uses it.
Full article: Who should own AI in a company your size?
- What should we manage centrally about AI, and what should we leave to each team?
Five things about AI are decided once, for the whole company: which assistant it pays for, the ground rules, the one plan, the order in which workflows get built, and who owns each workflow by name. Teams decide three things: which of their own work to propose, the detail of each workflow, and the checking of what each workflow prepares. Central decisions are made once and apply to every team, and the people who do the work make the rest of the decisions inside them.
Full article: What should we centralize about AI, and what should we leave to teams?
- Does a small business need a written AI policy?
Yes, a small company needs a written AI policy, and two pages is enough. It should name the approved assistant, say what information may and may not go into it, require a person to check and decide before anything is sent, paid or changed, and say who to ask. Banning AI until the policy exists only pushes people onto personal accounts the company cannot see. A policy says what not to do, so the company also needs a plan that says what to do.
Full article: Does a small company need an AI policy?
Rolling it out, and telling whether it works.
Where to begin, how to roll it out and how to tell whether it works.
- Which team should go first with AI?
Pick the first team for AI by three tests: its weekly work repeats more than any other team’s, that work begins from records someone can already export from a system, and its lead will own the first workflow by name. Enthusiasm and urgency do not decide it. If the obvious candidate fails one of those three tests, that team waits, and a team that passes all three goes first.
Full article: Which team should go first with AI?
- How do we decide which AI use cases to build first?
Choose the first three AI workflows from one team’s list, keeping only work that happens every week, whose steps can be written down in order, that one named person already checks, and that needs nothing more from business systems than a file someone already exports. From the work that passes, pick three that start from the same records and go to the same checker. Build them in order, and put the first into daily use before building the second.
Full article: How to choose the first three things to build with AI.
- How do we roll out ChatGPT or Copilot across the company?
A company rolls out ChatGPT, Claude, Copilot or Gemini by rolling out pieces of work, not licenses. Each team in turn gets one named piece of its own work that the assistant prepares, with written instructions, one owner and one person who checks the result. People keep using the assistant because it does their real work the way their best colleague would, and the next team begins once the first piece of work is in daily use.
Full article: How to roll out ChatGPT, Claude, Copilot or Gemini across a company.
- How do we get people to actually use AI at work?
Adoption of AI at work follows usefulness: people keep using it once it does a named piece of their own work the way the team’s best person does it, and once they see their manager using it too. Training sessions and encouragement rarely change daily habits; a workflow built from the team’s own method does. Involve the people who do the work in writing it, start with the task they like least, make checking the output part of the job, and never tie AI to monitoring or rating anyone.
Full article: How do we get people to actually use AI at work?
- How do we measure whether AI is paying back?
To tell whether AI is paying back, measure one whole piece of work against a baseline written down before the workflow first runs: how long the work takes from start to finish, how often it comes back with a mistake, and how long it waits. Record the same things once the workflow is in daily use, including the time a person spends checking. Active seats, hours of use and staff surveys do not show payback; if checking eats the time saved, fix the workflow or stop it.
Full article: How to tell whether AI is paying back.
The options, and what they cost.
The choices in front of you, and what they cost.
- Is there software for planning how a company uses AI across its teams?
Yes. Planning how a company uses AI across its teams can be done in software built for it. Many companies begin with a spreadsheet listing each piece of work with its team, owner and status, which works until the plan spans several teams and the most important work belongs to more than one. FullyPowered AI Planner is software built for the job: it holds a company’s teams, their roles and the work that passes between them in one plan, and the company chooses what to build first.
Full article: Is there software for planning how a company uses AI?
- Should we buy another AI tool or use what we already have?
Use the AI assistant the company already pays for first. Before buying another AI tool, name the piece of work it is meant for, test whether ChatGPT, Claude, Copilot or Gemini prepares that work well with written instructions, and buy only when the plan shows a gap the assistant cannot close. A specialist product earns its place when the work depends on data only that product holds or needs something an assistant cannot do; a purchase with no named work behind it becomes another unused seat.
Full article: Should we buy another AI tool or use what we have?
- Should our company standardize on one AI assistant, ChatGPT, Claude, Copilot or Gemini?
Yes, a company should approve one AI assistant as its default, because one company account means one set of rules, one place for shared workflows and one bill. Which of ChatGPT, Claude, Copilot or Gemini it picks matters less than having a default at all, so the sensible choice is the assistant already available through the software the company runs and already opened by its people. Workflows written as plain instructions move with the company if it changes assistants later.
Full article: Should we standardize on ChatGPT, Claude, Copilot or Gemini?
- Should we hire an AI consultant, and what does it cost?
A company does not need an AI consultant to start putting AI to work. Outside help earns its cost when the work is particular to the company and nobody inside has time to write it down, build the workflow and test it on real work. Before hiring anyone, write a one-page plan, pick one piece of work and try a workflow template on it, then weigh doing it yourselves, hiring a developer, training the team or bringing in help by the time each takes and what it leaves behind.
Full article: Should we hire an AI consultant, and what does it cost?
- What does it cost to put AI to work in a small company, for setup and each month?
With FullyPowered AI, setup is a fixed price known before work starts: FullyPowered AI Planner from $49 a year to plan it yourselves, the 3-Day Workflow Sprint at $750 to build one workflow from a written procedure, or the 2-Week Workflow Sprint at $3,500 to capture one team’s work and build its priority workflow. After setup, the workflows run in the AI subscription the company already pays for, so the running cost is that subscription and the team’s time to run and review the work.
Full article: What does it cost to put AI to work in your company?
How to start.
What to do this week.
- I run a company and feel behind on AI. Where do I start?
An owner who feels behind on AI starts with one decision rather than a hundred. This week, write a one-page plan naming the company’s teams, the first piece of repeat work each team moves onto AI, and the person who owns it. This month, build one workflow with one team; this quarter, put it into daily use and pick the next one from the page. The method is plan it, develop the workflows, deploy them.
Full article: I run the company and I feel behind on AI. Where do I start?
- How do we go from typing prompts into ChatGPT to AI work the whole team does the same way?
Every piece of work done with AI moves through four stages: Ask, Save, Share and Sustain. At Ask, someone types the request into the chat box each time. At Save, it is written down once so it runs the same way every time; at Share, the whole team runs that same version; at Sustain, one person owns it, updates it and checks it still works. Most work sits at Ask, and the value of the AI a company pays for grows at each stage above it.
Full article: Ask, Save, Share, Sustain: the four stages a piece of work goes through with AI.
- What should I do first in FullyPowered AI Planner during the trial?
FullyPowered AI Planner begins with a seven-day trial, everything unlocked: the trial starts with a card, nothing is charged for seven days, and you can cancel before then. Use the week to lay out every team in the company with the five or so workflows each should run with AI, invite the people who do the work to add what the suggestions missed, and mark one to three workflows to build first, each with a responsible team and a date.
Full article: Your first seven days in the Planner.
- What do we tell employees about AI when the company starts a plan?
Employees need to hear four things about AI, in one announcement from the person who runs the company: what is decided, what each team is asked to do, what stays the same, and where to bring ideas. The announcement goes out on the day the AI plan names its first team and first workflow, and every team lead repeats it to their own team that same week. It marks the move from experimenting with AI to using it on purpose.
Full article: What do we tell employees about AI?
- We have a plan. How do we build our first AI workflow?
The first AI workflow starts from one piece of work on the AI plan: write down how the company’s best person does it, precisely enough that an AI assistant such as ChatGPT, Claude, Copilot or Gemini prepares it the same way every time. The workflow sets out its inputs, steps, checks, output and the point where a person decides. Test it on last month’s real cases, then hand it to one named owner who runs it with the team.
Full article: We have a plan. How do we build the first AI workflow?