TL;DR Corporate leaders are restructuring their organisations around AI using a compression thesis: same work, fewer people, flatter hierarchy. The evidence increasingly points to expansion: more work, different work, new capabilities. Restructuring for compression when the actual phenomenon is expansion is a classification error with cascading consequences. Before redesigning your org chart, classify the change.

Your CEO just sent an all-hands email. The subject line contains the words "AI-first" and "restructuring." The deck references Block, Shopify, and Anthropic. There is a slide about flattening hierarchy, a slide about smaller teams, and a slide about everyone becoming a builder. The stock price ticked up 3% on the announcement.

"Coachable" is doing real work in real investment decisions.

No one has agreed what it means.

Fifteen years of research, term sheet language, accelerator screening forms, and coaching sales decks. Press any two practitioners for a definition and you will get two different answers.

Tatiana Somià published the most rigorous recent attempt last year in Cogent Economics & Finance. Fifteen competencies, five areas, stitched from four different literatures. When her investor panel and her coach panel disagreed about what counted as coachability, she resolved it by weighting the investors most. The final scale includes items the coaches in her own study said were coaching outputs, not coachability markers.

That is not a rigour problem. It is a construct problem. When you build a definition by negotiating across raters with different interests, what you produce is a map of what investors want to see in founders. Useful. Not the same thing as coachability.

Here is the definition I am willing to defend.
Three capacities, one precondition.

• Seek feedback: the capacity to incorporate signals that do not originate in your own perception.

• Reflect: the capacity to think past the mental models that constructed your current view.

• Act: the capacity to break habits that worked at an earlier stage and no longer fit.

The precondition is courage. Before any of these can operate, the founder has to be willing to deliberately expose a blind spot. Not discover one by accident. Not admit one after the consequences are undeniable. Go looking for the view they have been avoiding.

Key takeaway: The pilot that converts into a contract is not designed to prove the technology works. It is designed to arm the internal champion with the evidence the organisation's governance process requires to say yes. Three criteria separate pilots that convert from pilots that demonstrate: representativeness, boundedness, and learnability.

TL;DR: Recent research identifies four distinct ways corporates absorb AI through startup partnerships. The technology matters far less than the collaboration design. When both sides are operating in different modes without knowing it, the partnership stalls. One honest conversation can fix this.

Corporate innovation programmes secure funding but skip internal go-to-market. BCG data shows 83% of companies call innovation a top-three priority, yet only 3% have the operational readiness to act on it. The gap is not strategy. It is adoption: who will change their daily work, what it will cost them, and why the initiative team has not answered that question.

Image Source: Dorian Darko | https://replicate.com/doriandarko

Most startup sales failures are not sales failures. They are positioning, ICP, or validation problems that surface at the point of sale. Every major sales methodology, from Challenger to MEDDIC to Sandler, assumes upstream work that most founders have not done. Fixing your sales process when the real problem is two layers upstream is the most expensive way to learn nothing.

TLDR;-) Large organisations don't resist innovation because they're broken. They resist it because they're working. Four mechanisms do the killing: procedural resistance, resource competing, standard dilution, passive waiting. Each is a legitimate organisational function operating in the wrong context. If the immune response is firing, you're probably working on the wrong layer.

Eighty-three per cent of companies rank innovation as a top-three priority. Three per cent are ready to act on it. That is not a typo. BCG's 2024 Most Innovative Companies report calls what remains "zombie innovation systems": organisations going through the motions of innovation without strategic commitment, waiting for certainty that will never arrive.

The Critical Path Layers framework identifies four cross-cutting dynamics that accelerate or impede corporate innovation at every stage: political capital, clock speed, organisational immune response, and the narrative layer. The first three are constraints. The narrative layer is different — it's the medium through which the other three operate. Political capital is spent and earned through stories about the initiative. Clock speed mismatches become visible when the story stops evolving. The immune response deploys counter-narratives to neutralise change.

What makes the narrative layer structural rather than cosmetic is that it must evolve as the initiative progresses through the CPL's layers. Each layer produces a different story because the initiative's relationship to the organisation has changed.

AI Innovation!?
The failure is not in the technology. It is not in the strategy. It is in the classification: the pre-strategic decision about what kind of problem this is.

Every corporate AI initiative has two identities. What the organisation approved, and what the initiative actually requires. The gap between these two is the misclassification, and it cascades through every downstream decision.

AI for VC & Founders: The playbook (dealbook?) has changed—and everyone is scrambling to keep up.