In short
- Four signatures reveal an AI-suited process: gathering context, handovers, triage and synthesis.
- Where the work is mostly about interests and judgement, the human stays indispensable.
- The useful question: which of those four carry your worst Mondays?
The pattern behind our own choices
When we redesigned our own operation, we didn't choose at random. The processes we picked all had the same shape. We see that shape recur at clients, across sectors. Four signatures reveal a process that lends itself to AI.
1. Gathering context together
Work where someone first fetches information from everywhere before the actual task begins. At a builders' merchant we saw a customer-service team look up delivery times every day by searching two systems that don't talk to each other. The context was there, just scattered. That costs minutes per question, and multiplied by a day of questions you see where the time goes. This type of work is the clearest starting point.
The question is not where you can deploy AI. It's which of these four forms carry your worst Mondays.
2. Handovers where information falls away
Processes with many links, where at each handover something is manually re-entered or re-estimated. At the same logistics operation, an order travelled through a chain of export to Excel, manual counting, a physical walk-round, and back into the system before a route existed. Every step is human work that is mostly retyping and transferring. That is exactly where there is room.
3. A stream that needs triaging
An inbox, a ticket queue, a stack of incoming questions that first need sorting and contextualising before someone picks them up. Our own Monday email is an example: out of everything that is at play, an agent filters what needs a decision. For an insurer or a service department, that is the same movement, just on claims or customer questions.
4. Synthesis and reporting
Work where separate sources come together into a text: a report, a summary, a write-up. In advisory work a lot of time goes into this. Generative AI is strong at it, provided a human makes the final call. It saves the first draft, not the judgement.
Where AI stays limited
In a strategy session with an advisory firm it came out sharply. A participant noted that most of their value doesn't sit in the data, but in navigating interests between parties: politics, governance, delivery context. Several people agreed. That is work where AI adds little for now, and it is not a weakness to repair. It is a strength to protect. The art is in the distinction: automate the looking-up and the retyping, guard the judging and the translating.
The question you ask yourself
So the question isn't so much where you can deploy AI. Almost everywhere something is possible. The useful question is which of these four signatures carry your worst Mondays. Where does someone gather context every day? Where does information fall away at a handover? That is where it begins, whatever sector you're in.
Ruben Horbach
Co-founder · Back From the Future
Ruben researches how organisations adopt AI meaningfully — not as technology, but as a change in work and people. He builds the agent infrastructure behind BFF and speaks about the near future of work.
