Stop experimenting. Start using.
We encounter this on almost every project. We walk into a company, meet busy, capable people — and in the course of process mapping or interviews, it becomes clear that most of them spend a significant part of their day on tasks that could be done in a fraction of the time. They manually copy data from one system to another, read 30-page documents to find three relevant paragraphs, and prepare reports that look the same every time — just with different numbers.
It is not that these people are inefficient. It is that no one has yet shown them specifically how AI can help with their actual work. Honestly, what I am writing now is something I would not have written six months ago. Even we are surprised by what AI tools can do today — and we share new tips and use cases for everyday work every single day.

Meetings and their outputs
A classic example, one that many people are already aware of. A manager who runs four to five meetings a day and writes the minutes from memory in the evening. We show them how Copilot in Teams automatically transcribes and summarises every meeting — including task assignments and key decisions. They try it with mild scepticism. The next day they message us to say they saved an hour and a half. This is not an exception — it is the new standard.
We also see a remarkable shift in data analysis. When you feed data into Claude AI correctly and know how to ask the right questions, the outputs are extraordinary — and we are not just talking about basic analysis, but robust statistical models. The outputs from Claude are visually clean, requiring no further editing — simply ready to use.

Data analysis without statistical expertise
Another situation we see regularly: an analyst or project manager who spends half a day each month building a project plan or status report in Excel. Formulas, pivot tables, charts — all done manually. When we show them how to upload a spreadsheet into Claude or Copilot and simply ask — where are the anomalies, what trends do you see in this data? — and receive a finished analysis with charts in under two minutes — there is usually a moment of silence. Then comes the question: Does this work with our data? Yes, it does.
Copilot can now edit data directly in Excel — there is no need to upload anything elsewhere, everything happens in the tool people already know. That is a genuine revolution. Frankly, training people on complex Excel functions is becoming somewhat redundant. A personal example: I use XLOOKUP regularly but can never remember its parameters — now I just describe what I want, and it is done.

Working with company documents
This is something few people are aware of — and once they try it, they never go back. You upload company documents directly into Claude — manuals, reports, meeting notes — and then simply ask questions about them. No scrolling through dozens of pages. Just a question and an answer. Claude handles long documents better than most other tools, making it the ideal choice for working with extensive company materials.
In our training sessions, we demonstrate this using the participants’ own documents, which they bring with them.

Automating repetitive processes
One team faced a situation where every incoming customer request had to be manually entered into their internal system, a notification sent to a colleague, and a task created in Trello or Planner. Three steps, five minutes, twenty times a day. Using Make, we connected everything in a single automation — and the entire process has run by itself ever since. We later discovered that the same result could be achieved even faster directly in Claude. In our training, we cover all of this hands-on — working through a series of real-world examples together.

Building your own AI agent
This is an area that has accelerated significantly over the past year. Using Claude, ChatGPT, or Microsoft Copilot Studio, anyone today — without any programming knowledge — can build their own AI agent. At ICG, we primarily use Claude as our main tool — for its ability to handle long texts, the accuracy of its outputs, and its natural way of communicating.
In our training, we build agents together: participants arrive with a specific idea of what would make their work easier, and leave with a working agent. An onboarding assistant that answers common questions from new colleagues. A bot that checks documents against a company template. An agent that prepares summaries of customer feedback. This is not the future — companies are doing this right now. And the course certificate you receive will also be prepared by an AI agent.

Where to start if you want to develop your AI skills?
The biggest barrier is not the technology. It is the mindset. The people who get the most out of AI do not think about what AI can do. They think about what they do repeatedly, what takes too long — and then try whether AI can help. This is exactly what we aim to pass on in our three-day training programme Digital Techniques and AI Tools — from the fundamentals of prompting and working with Claude, through data analysis, to building your own agents and no-code applications. Everyone leaves with a working tool built for their own job and an action plan for moving forward.

What does it actually deliver?
For the past two years, our process audits have actively identified opportunities for automation and digitalisation. We work with internal teams to put numbers to our recommendations. Here are a few examples:
– A production planning team reduced working hours by 5,000 per year following the introduction of AI and RPA
– A chatbot at a customer service centre handles 40% of incoming questions and simple requests
– An AI assistant reduces report preparation time by an average of 50–70%
– Automating repetitive processes saves a ten-person team more than one full-time equivalent per week

Want to learn more? Visit our Digital Techniques and AI Tools training or contact us at office@integratedconsulting.cz.