AI productivity with ChatGPT or another language model doesn’t mean leaving everything to a tool; it means handing over specific, recurring tasks to it. You spend the rest of your time on work where your judgement makes the difference.
Below are six practical ways to do that, from meetings to documents. The first five are about saving time. The sixth covers something missing from most lists: how to stop the time you save from leading to a mistake that costs you more time afterwards.
None of these approaches requires technical knowledge. They work with the standard interface of ChatGPT, Claude or a similar model, without extra software.
If you want to apply these six approaches systematically rather than picking out individual tips, LearnLLM’s ChatGPT Course builds a fixed way of working around them.
Summarise meetings instead of taking minutes
Summarising meetings with AI saves the most time for meetings you already record or take rough notes of. Paste your notes or the transcript into the model and ask for a summary with action points, who is responsible for what and which decisions were made.
Ask specifically for decisions and open questions to be kept separate. Without that instruction, you often get a flat list in which the two are mixed up, while that distinction is exactly what determines what needs to happen after the meeting.
Draft emails and messages faster
Drafting emails faster with AI works best when you supply the core of the message yourself and let the model help with the wording, not the other way round. Say who will read the email, what you want to achieve and what tone to use.
For recurring situations, such as replying to a complaint or confirming an appointment, it pays to keep a fixed structure that you reuse every time.
Summarise long documents before you read them
Summarising long documents before you read them gives you a framework for working through the rest more quickly. As well as a summary, ask for the main risks or deviations from what you expect.
How much background you provide matters more here than the length of your prompt. A summary for yourself can be different from a summary you forward to a colleague who doesn’t know the context.
Save recurring tasks as a fixed prompt
Saving recurring tasks as a fixed prompt stops you having to work out each time how best to ask for something. If, for example, you update a report every week, keep the prompt that works for it and only change the figures or text.
Most tools let you save such prompts within a project or workspace, so your colleagues can use the same approach instead of each reinventing the wheel.
Start small with one task instead of everything at once
Starting small with one task achieves more than trying to reorganise your whole working day in one go. Choose the task you do most often and that varies least, because that is where the effect shows fastest.
Only move on to the next task once that one runs smoothly. If you try everything at once, you often lose more time experimenting than you eventually save.
Always check the output before you use it
Checking the output before you use it is the step that turns time saved into real time saved, instead of a risk you have to put right later. A language model presents a wrong answer just as convincingly as a right one, and you can’t tell the difference from the tone or layout.
At the very least, check figures, names and references to sources or regulations, because those are where a model most often hallucinates. That check takes a few minutes and stops you paying back the time you just saved, and more, on a correction afterwards.
Does AI replace your own skills?
No, AI doesn’t replace your own skills; it shifts what you use those skills for. Assessing an answer, spotting a mistake and making a decision remain human work, even if drafting a first version goes faster.
How much AI productivity can you realistically gain?
How much time you can realistically save depends heavily on how repetitive your work is and how consistently you use the tools.
In a controlled experiment with 453 professionals (Noy and Zhang, Science, 2023), short, well-defined writing tasks (20 to 30 minutes) took 40% less time with AI, with 18% higher quality. A percentage like that applies to the part of your work AI is good at, not to your whole working day.
How do you make your whole team work smarter, not just yourself?
Your whole team only works smarter in a lasting way if everyone uses the same way of working and the same check, instead of each employee inventing their own approach. Without that alignment, you end up with as many different ways of working as there are employees, with varying quality as a result.
Frequently asked questions about AI productivity at work
Working smarter with AI often raises practical questions in the workplace about rules and privacy.
Are you allowed to use AI at work?
You may use AI at work if your employer allows it; the EU AI Act doesn’t prohibit ordinary office tasks with an AI tool. Under Article 4 of the AI Act, organisations that use AI do have to take measures that support the AI literacy of their staff. So check whether there is an internal AI policy, and only put customers’ personal data into a tool once your organisation has made arrangements for it.
Can you use the free version of ChatGPT for work?
You can use the free version of ChatGPT for work, but OpenAI trains its models on your conversations as long as the “Improve the model for everyone” setting is switched on. You switch that setting off under Settings, in Data Controls. With ChatGPT Business and Enterprise, OpenAI doesn’t use the organisation’s data for training by default. So don’t put confidential client information into a personal free account.
Can you use ChatGPT in languages other than English?
Yes, you can set ChatGPT to languages other than English, because OpenAI lists a range of supported interface languages, including Dutch. You choose the language in your account settings. Separately from that setting, you can write your prompts in your own language and ask for an answer in that language, even if the interface is in English.
How do you learn to apply this systematically instead of experimenting?
You learn to apply this systematically by going beyond individual tips and building a fixed way of working around the tasks that deliver the most for your work.
You build that way of working step by step, with the check from this article as a fixed part rather than as a separate warning afterwards.
