AI and AI agents invite us to play. Knowledge can be “mined” – in seconds and for mere cents. So it's no surprise that the member companies of the Advanced Foresight Group are increasingly seeing client requests like these.
- “Can you build us an AI agent workflow pilot? Going forward, we want to manage without foresight service providers, and we'd like to benchmark different solutions first.”
- “Can you build us an instant radar? We've seen that you can do it!”
- “Even the British minister is doing foresight himself these days. We want that too. Do you have copy-and-paste AI workflows for n8n or Gumloop that we can adopt?”
The answer, of course, is: yes, we can build that. But is that really enough?
Cutting spending on foresight is currently in vogue – a very dangerous tightrope walk for providers and organizations alike. In-house strategic foresight often fails to realize that it is clipping its own wings.
Instead of strengthening their internal position, people build tools. Instead of gathering learning experiences in shared processes, they celebrate their own self-efficacy. I remember a project in which the mayor said he could paint images of the future himself. So he started drawing in MS Paint – much to the delight of his entourage. Let's not make the same mistakes! “The Emperor's New Clothes” should teach us that the arrogance of thinking something is “so easy” leads nowhere.
Just because new tools “would enable anyone” doesn't mean the foresight processes are automatically of high quality. What's missing is workflow design and a corrective. What's missing is friction, participation, expert knowledge, and the careful design of images of the future.
More and more often, you stumble across posts like this on LinkedIn:
“I built a blog-writing AI foresight workflow that runs in 33 seconds.
It takes a foresight topic I give it and:
→ pulls research using Perplexity
→ references my brand voice + writing style
→ generates a full blog post with structure + citations
→ sends it to me in Google Docs + Slack
This isn’t just pasting a prompt into ChatGPT.
The workflow pulls from a custom brand doc — my tone, formatting preferences, and real writing samples — so the output actually sounds like me.
It doesn’t just generate content.
It generates my content, automatically.
Built the whole thing in Gumloop — no code needed.”
👇 Comment “Workflow” if you want the full tutorial on how I built this.
AI agents will always produce knowledge and generate volume. But it is precisely this “generated” volume that clouds our view. It lures us in: “Don't be stupid and do the work yourself when I can do it for you. I'm fast, cheap, and self-explanatory – just copy me.”
The long-term result: brainless organizations! Triggered by workflows, they drift along without a leader. Because foresight, ideation, communication, and strategic direction are inseparable. Not “brainless and dumb as fuck!”
AI-supported processes need clear quality standards so the team doesn't bask in cognitive offloading – without noticing that it is making itself more replaceable. You gain a competitive edge when you experiment. There's nothing wrong with setting up pilots. Foresight budgets should regularly challenge foresight service providers. But please take these pilots seriously!
If you need real support with pilots like these, talk to us!
We know our stuff. And we're happy to tell you more about how you
- create real added value and strengthen your resilience,
- rediscover the joy of shaping things, and
- get your hands really dirty so that something new can emerge!
So go for it! First the pilot, so you don't build up dependencies – and then the transformation, together.
