9 things organisations need to plan for AI to become part of everyday work

By now, many organisations have given employees access to generative AI productivity tools such as Microsoft Copilot, ChatGPT, Gemini and Claude.

They may also have delivered training, published internal guidance and started measuring how often the tools are being used, so on paper quite a lot appears to be happening. The question is whether anyone has actually planned how all of those pieces are supposed to help people change the way they do their jobs.

Licences, training, guidance, management involvement, practice and measurement all affect what happens next.

When they are treated as separate activities, an organisation can see growing AI usage without knowing whether employees are saving time, improving the quality of what they produce or reaching a more consistent standard.

That is often a planning mistake.

The aim is for AI to become a normal part of completing the tasks where it genuinely helps.

People need to understand where it fits into their day, how to work with it effectively, which tools they are allowed to use, what needs checking and how to repeat a good result when the task changes.

There are nine areas worth planning for.

1. Decide which parts of people's jobs AI should help with

One of the first things an organisation needs to understand is where AI could genuinely help employees during their normal working day.

This sounds obvious, although it is still common to begin with licences, training dates, features and tool demonstrations before anyone has properly looked at the tasks people spend their time completing.

• A sales team may spend significant time preparing for customer meetings, researching accounts, reviewing email conversations, writing follow-up communication and creating account updates.
• A procurement team may review proposals, compare suppliers, prepare scopes of work, analyse documents and bring information together to support a decision.
• An HR team may prepare employee communications, review feedback, draft documents, summarise information and prepare management updates.

Those examples give the organisation something specific to examine. How long does the task take today? What information is needed? What does a good result look like? Which parts require judgement? Where could AI save time? Where could it improve the result? What still needs to be checked by a person?

Once those questions have been answered, it becomes much easier to decide what people should learn and what success should look like.

If employees cannot see where AI fits into their day, knowing a long list of features won't help much.

2. Be clear about what the training is supposed to achieve

A one-day or two-day AI course can achieve a lot when the objective is clear.

A general session can help people understand what generative AI is, what the main tools can do and some of the areas where they may help. For a mixed audience, that may be exactly what is needed because the priority is awareness and giving people enough understanding to start exploring.

A focused session with one department can go much deeper.

Participants can spend the time on a smaller number of tasks they genuinely need to complete, practise those tasks properly, review the results and improve how they are working with AI.

Someone in Procurement may leave knowing how to review a proposal, compare information from several suppliers or prepare the first draft of a scope of work using a clear process they can repeat.

Problems begin when one or two days are expected to cover everything.

Participants may be shown several tools, dozens of features, prompting techniques, security guidance, business examples and a long list of possible uses. They can leave impressed by what they have seen and still be unsure what they should do when they sit down at their desk the following morning.

If the aim is awareness, design the training around awareness.

If the aim is to help a team become very good at a small number of important tasks, spend the time on those tasks and teach them properly.

Trying to cover too much usually leaves people knowing a little about many things and very little in enough depth to change how they do their jobs.

3. Teach people a clear way to work with AI

Seeing where AI can help gives people a reason to use it. They also need a clear way of working through a task with it.

We see this frequently in training.

Someone opens an AI tool and immediately gives it an instruction such as “write this”, “summarise this”, “analyse this” or “create a presentation”. The tool produces something and the employee decides almost immediately whether the response looks good.

Sometimes it is useful. Sometimes it is generic, incomplete or based on a poor understanding of what the employee actually wanted.

One simple approach we teach is to use the first interaction to explain the task before asking AI to start completing it. Explain what you are trying to achieve, give it the relevant information, explain who the output is for and what the result needs to contain, then ask AI to tell you what it understands the task to be.

If the understanding is wrong, correct it before continuing.

That may only take a few minutes, but it gives everything that follows a much stronger starting point. The employee can then work through the task in stages, provide additional information, review each result, ask questions, correct problems and improve the final output.

Many important business tasks need several interactions. The skill is knowing how to guide AI through the task and what to do when the first answer is not good enough.

That way of working can then be used again across different tasks and different AI tools.

4. Plan what happens after the training

People can learn a lot during a training day, particularly when they have spent the time practising tasks that are relevant to their jobs. The difficulty typically comes afterwards, when they return to the pace of their normal working day: meetings, emails, deadlines and the routines and habits they have been using for years.

Someone may try an approach from the course, get a poor result and have no idea whether the problem was the information they provided, the instruction they gave or the way they structured the task.

Another person may get a good result, become very confident and start trusting AI more than they should. Someone else may fully intend to practise during the week and simply never get around to it.

This is why practice after training needs to be planned.

People need time to use what they have learned on their own tasks, because that is when different questions come up. The source information may be less structured than it was in training, the task may be slightly different from the training example, AI may misunderstand something or the employee may realise that the task needs several stages.

They may produce something that looks good and then discover that a few important points are missing.

Those experiences give people something real to learn from.

Training programmes that include follow up clinics give people a chance to bring those problems back, review what happened, correct the approach and try again. With repeated use and feedback, people become better at deciding what to do themselves and gradually need less support.

A one-day or two-day course can build awareness and teach valuable skills. Changing how hundreds of people use AI during their normal jobs requires continued practice and reinforcement.

That could mean another workshop, a short clinic, time between sessions to try specific tasks, support from managers or regular discussions where employees can bring examples and questions.

People need another opportunity to practise, ask questions and improve after the course finishes.

5. Give employees clear guidance on which tools they can use

Employees need clear answers to three simple questions:

Which AI tools can I use? Which ones can I not use? Why?

An employee may have Microsoft 365 Copilot through the company, a personal ChatGPT account, access to Gemini and colleagues recommending Claude. Without clear guidance, people start making their own decisions about which tools to use and what information they can put into them.

Some become overly cautious and avoid useful tools completely. Others may use a tool in a way the organisation would never have approved.

A policy may contain everything the organisation needs from a legal, security or compliance point of view. Employees still need to understand what that means when they are doing their jobs.

If company information can be used in one approved environment, explain that clearly. If certain information cannot be entered into another tool, explain that too. If one tool is approved for a particular type of task, employees should understand why and they should know when a particular type of information needs extra care.

The guidance needs to be clear enough that someone can read it, understand it and know what to do.

6. Keep reinforcing AI use internally

Training may create a lot of interest for a few days or weeks. Then normal business pressure takes over, people return to familiar habits and much of what they learned moves further down the list of things they are thinking about.

This is why internal communication matters so much.

Share examples of tasks that worked well, show where a team saved time, explain where an AI result needed correction and give people short examples they can try themselves. Different departments should be able to see how colleagues are using AI in situations that feel relevant to them.

If someone in Procurement finds a good way to review a certain type of document, share it. If a manager uses AI to prepare a weekly update in half the time, show people how. If a team discovers that a particular approach produces poor results, that is also worth sharing because it helps other people avoid the same problem.

Examples from colleagues can be particularly effective because employees can see how AI is being used by people doing similar jobs.

The message also needs repetition. People will not remember everything they saw during training, new employees will join, tools will change and new tasks will appear.

The organisation has to keep showing people what good AI use looks like.

7. Make managers responsible for keeping AI in the conversation

L&D has an important role in helping people learn, technology teams have an important role in tools, access and security and Heads of AI usually coordinate activity across the organisation. Managers also have a major responsibility because they are closest to the tasks their teams complete every day and can see where people are struggling or finding value.

AI should therefore become a regular part of team, department and management conversations.

It does not need an hour in every meeting. Ten focused minutes can be enough to ask what people used AI for that week, where it helped, where the result needed too much correction, which task the team should try next and what guidance is still unclear.

These conversations help people learn from one another and keep AI connected to the things the team actually needs to get done. They also give managers a clearer picture of what is happening instead of relying only on usage figures or feedback collected immediately after training.

Important points can then feed into a leadership level AI council. Leaders can see which teams are getting results, which tasks are proving useful, where employees are struggling, whether there are common questions about security or approved tools, and where more training or support is needed.

The team level and leadership level should connect.

If AI only appears in training calendars, technology updates or occasional internal announcements, management responsibility is missing.

8. Measure whether people's jobs are improving

Most organisations can and do measure usage. They can see licence activation, active users, number of interactions and changes in usage over time, which gives them a view of how often the tools are being used.

The organisation also needs to know what has changed in people's jobs.

Is a task taking less time? Has the quality improved? Are people producing a more consistent result? Can employees complete tasks that previously required a lot of effort? Are they completing more of the right tasks? Has their confidence improved? Can they repeat the approach without regular support?

Imagine one employee uses Copilot fifty times during the week and produces very little of value, while another uses it three times, saves four hours and produces a better management report.

The second employee may be creating far more value.

Start with a small number of important tasks and understand how they are completed today. Then look at what changes when AI is introduced. Time saved, quality improvement, consistency, tasks completed and employee confidence can all help build a clearer picture.

If usage is the main measure, the organisation still does not know whether its AI investment is improving anything that matters. Usage alone is a vanity metric. Focus on time saved, quality, consistency and the tasks people are completing better.

9. Check whether the improvement can be repeated

One good result is encouraging. The real test comes when the employee has to do something similar again and the information, audience or task has changed.

Can they adjust what they are doing and still reach a good result?

Someone may learn an excellent prompt during a course and use it successfully the next day. That gives them one useful example. Over time they need to understand why the approach worked so they can adjust it when the circumstances change.

They should know how to define the task, provide the right information, explain the result they need, work through several stages, check what AI produces and improve the output. When they can use the same thinking on a different task, the improvement starts to become something they can rely on.

This also reduces dependence on a small number of enthusiastic employees who happen to be good at using AI.

Have we actually planned this properly?

Look across the nine areas together.

Do people know which parts of their jobs AI can help with? Is the training designed around a clear goal? Are employees learning a way of working they can use again? What happens when the training finishes? Do people know which tools they can use and why?

Is the organisation continuing to show people practical examples? Are managers discussing AI with their teams? Are you measuring changes in time, quality and consistency? Can employees repeat the improvement when the task changes?

If several of those answers are unclear, there is a gap in the plan.

The organisation may have bought or approved the technology, delivered training and started measuring usage without fully planning how all of those pieces are going to help people improve how they do their jobs.

The aim is for people to use AI naturally in the parts of their jobs where it saves time, improves the result and helps them maintain a stronger standard.

When that starts happening regularly across teams, AI is becoming part of the normal working day.