Using AI to rethink the work, not just speed it up

Asked to automate the production of client deliverables with AI, I advised stepping back first: mapping the outcome they actually wanted rather than the existing process, so AI could open up new ways of working instead of just speeding up the old ones.

As the engagement continued, Orbit Partners asked me how they could use AI in their work, starting with automating the production of client deliverables.

They came with concrete, sensible ideas: specific points in the existing process where AI could save time. My advice was to step back before building any of them. It is tempting to point AI at a process and make it faster, and that is often worth doing, but the bigger value comes from asking what you are actually trying to achieve, separate from how the work happens to be done today.

So rather than start from the mechanics of the current process, we mapped it from the outcome: what a good client deliverable actually needs to do, and what would genuinely move that forward. Framing it that way opens the door to doing things in a new way, instead of quietly baking today's workflow, and its accidental constraints, into an automation that is then hard to change.

That is where AI advice is worth most: not "which steps can we automate?", but "now that this is possible, what should the work look like at all?"