Every week, somebody offers a university commercialisation office an AI tool. A scouting platform, a triage copilot, a diligence assistant, a pitch generator. The pitches share one assumption: that everyone agrees what the work actually is. Buy the tool, plug it into the workflow, save the time.
Here is the problem with that assumption. Nobody has mapped the workflow. Not properly, not from the inside, not in a form that a busy office would recognise as its own week. The sector holds decades of process diagrams drawn for annual reports and funder submissions, and almost nothing that describes how an opportunity really moves from a corridor conversation to the first external commitment, where it stalls, who owns it while it stalls, and what improvised tooling, the spreadsheets, the workarounds, the one colleague who remembers everything, actually carries it.
That gap matters more than any individual tool. Assess tools against an imagined workflow and the assessment is fiction with a scoring rubric. Assess them against the real one and something useful happens: half the pitches disqualify themselves, because they solve stages that were never the constraint, and the genuine jams stand out, usually unglamorous, usually somewhere around ownership and routing rather than anywhere a demo looks good.
Assess tools against an imagined workflow and the assessment is fiction with a scoring rubric.
So before this practice assesses anything, we are doing the unfashionable part first. We have drawn the workflow as we know it from the build side: seven stages, from opportunity capture to first external commitment, with the branch made visible at the route-to-value stage, because a licence, a partnership, a venture and a considered stop are all legitimate destinations, and any map that treats spinout as the default is describing an investment thesis, not a commercialisation system.
And then we are doing the part that makes it evidence rather than opinion: publishing it and asking the people who live it to mark it wrong. Which stages are missing. Where the work really jams. What your office actually uses at each stage, commercial, internal, home-grown or held together by one heroic spreadsheet. The corrections come back to contributors first, as a corrected map and a vendor-neutral view of the tools and initiatives genuinely in use, with practical pros and cons rather than anyone's marketing.
This is the practice's founding conviction applied to its newest question: evidence built with the system, not about it. The map is only as good as the corrections, and the corrections only come from you.