Blog
AI agents and automation
An AI agent carries out steps in your systems by itself: it reads email, looks up information, fills in a form or starts the next task. A regular chatbot only answers. That makes agents useful for recurring work, but it also calls for clear limits: what the agent may see, what it may do and who approves it. Here you can read what agents and skills are, when they pay off and how to start without taking unnecessary risks.
Start here
All articles (9)
30 October 20268 min readAgentic AI: when should an agent start on its own?
Agentic AI in practice: what can a proactive agent do alone, what must it propose and what never? A per-action framework with an owner, a log and a budget.
29 October 20268 min readClaude skills: write them once, use them in every AI tool
Claude skills follow an open standard that ChatGPT, Copilot and Gemini also read. How to write a skill once, give it an owner and install it in every tool.
21 October 20267 min readWhat is an AI agent? And when do you actually need one
What is an AI agent? A language model that picks its own steps in your systems. With dots, Gemini and Copilot Autopilot, and when a workflow fits better.
20 October 20267 min readAI agents in SMEs: is Europe falling behind?
AI for SMEs: OpenAI reports fast agent growth, official statistics measure something else. With a four-stage self-check and how to choose your first process.
11 October 20265 min readGemini agent: Google's AI colleague with its own email address
Google's Gemini agent works in Gmail, Docs and Sheets and gets its own account, audit log and spend cap. What it is, when it arrives and what to arrange first.
11 October 20264 min readGemini Gems are becoming skills: what do you need to do?
Gemini Gems are going away: Google is replacing them with skills. Business accounts stop no sooner than 1 March 2027. The dates and what to do with your Gems.
30 September 20265 min readAn AI chatbot that knows your company knowledge
An AI chatbot answers employees' questions from your own documents, in Microsoft Teams. See what that looks like, with HR as an example, and what it takes.
30 September 20266 min readOpenAI dots: what can you do with them in Europe?
Dots are OpenAI's always-on AI agents. Who gets them in Europe, what they cost, how much control you have and what to settle before a first trial.
28 September 20265 min readUsing AI agents safely: five questions before you start
Deploying an AI agent safely starts with AI governance: minimal permissions, an owner, a log and a human in the loop. With five questions before you start.
Frequently asked questions
- What is the difference between an AI agent and a chatbot?
- A chatbot answers questions. An AI agent also takes actions, such as sending an email, creating a document or updating data in a system, and can chain several steps together.
- What is agentic AI?
- Agentic AI is the umbrella term for AI systems that pursue a goal on their own: they plan steps, use tools and check their own result. How independently an agent may work is up to you.
- What is a skill?
- A skill is a written-down way of working, with instructions and optionally examples, that an AI tool uses whenever a task calls for it. You capture a way of working once and then use it the same way every time.
- How do you start safely with AI agents?
- Start with one well-defined process and give the agent access only to what it needs. Have a person approve actions with consequences, such as sending or paying, and decide who owns the agent.
Newsletter
The AI news that matters, once a month
One email a month with the most important developments in AI and what they mean for your organisation. No noise.
Let us begin
Want this in your own organisation?
We build it, or we teach your team to. Either works.