What we see
from inside.

A PowerPoint with boxes and lines. A belief that if you redraw the chart, you change how the company operates. Sometimes the chart changes three times in eighteen months. The company operates exactly the same way throughout.

Structure is the last thing you should change. It is also the easiest thing to change, which is why it gets changed first.

What we do before we touch any structure is simple: we map the actual decision-making. Not the theoretical one. We ask who made the last three major decisions in the company. We ask who was in the room, and who found out afterwards. We ask where things get stuck and how long they have been stuck there.

Ninety percent of the time, the answer is not that the structure is wrong. The answer is that the wrong people are in the wrong seats, accountability is unclear, or the leadership team has never agreed on what they actually own.

AI has changed this analysis considerably. We now run language models across email threads, meeting notes, and decision logs to surface the informal power map before we ever sit down with the leadership team. It surfaces patterns faster than months of interviews. It also surfaces things people are not comfortable saying out loud.

The org chart tells you the intended design. The data tells you the actual one. They are almost never the same.

What this means in practice

If your company is considering a restructure, ask this first: can every person in the leadership team describe, in one sentence, what they are personally accountable for delivering in the next twelve months? If the answers are inconsistent, or if people look at each other before answering, the structure is not your problem.

Fix accountability first. Then fix structure. In that order. Every time.

Companies that restructure without fixing accountability do not fix the problem. They move it to a different box on the chart.

Investors do not buy stories. They buy confidence that the story holds under pressure. The pitch is thirty minutes. The due diligence is sixty days. What kills most rounds is not what happens in the pitch room. It surfaces in the data room.

There are four questions we ask every founder before they enter a fundraise. The answers tell us everything.

First: Can you explain your unit economics in two minutes without a deck? Not the headline numbers. The mechanics. What does a customer actually cost you to acquire, serve, and retain, broken down by cohort?

Second: Is your cap table clean? Not good. Clean. Do all your shareholders know they are shareholders? Are there any side letters, informal agreements, or promises that have not been documented?

Third: What does your financial model assume about the next eighteen months, and what has to be true for those assumptions to hold? This question has a right answer. Most founders do not have it.

Fourth: If your lead investor asked tomorrow to speak directly with three of your customers, unsupervised, what would those customers say?

AI has changed how we do this preparation. We run models across financial data, customer records, and operational metrics to identify the questions an investor's analyst will surface before the investor does. We find the gaps before they become the reason the round stalls.

The moment most founders miss

The six months before a fundraise are more important than the six months of the fundraise. That is when the company is built into something fundable. Or it is not. The pitch is the last mile, not the race.

If you are thinking about raising in the next twelve months, start the preparation now. Not the deck. The company.

Cultural fluency is table stakes, not a differentiator. Everyone applying to be your first hire in Hong Kong speaks Cantonese and English and has worked for a Western company. That narrows the pool but it does not tell you who can actually build a business in a market where your brand is unknown, your product is unproven, and your headquarters is eight time zones away.

The first hire in a new market needs one quality above all others: the capacity to operate without infrastructure. No brand recognition. No established process. No colleague two desks over to ask. They are, functionally, a founder, with a salary and a boss who does not understand what they are dealing with.

We have seen this play out across the HK-Europe corridor more times than we can count. The hire with the impressive network and the right university who cannot make a decision without approval. The operator with the less polished CV who builds the first three customer relationships in month one and closes the first deal in month four.

AI has changed the hiring process for market entry significantly. We now run structured capability assessments (scenario-based, not personality tests) that identify autonomy, judgment under ambiguity, and commercial instinct before the interview stage. The data consistently surfaces candidates that would not have made the shortlist on CV alone.

The question to ask in every interview

Tell me about a time you had to make a significant decision with incomplete information and no one to defer to. What did you do, and what did you wish you had done differently?

The candidates who answer this well, specifically and honestly and with genuine reflection on what they got wrong, are the ones who will survive the first twelve months in a new market. The ones who tell you about a team success are the ones who will be back in your office in six months asking for more support.

The consulting industry spent two years debating whether AI would replace consultants. It asked the wrong question.

The right question was simpler: what were those consultants actually delivering that a well-structured prompt cannot?

For a significant portion of consulting work (the benchmarking, the framework application, the slide production, the market sizing) the answer is: not much. A capable AI model, well-directed, produces that work faster, cheaper, and without the coordination overhead of a team that needs to be managed.

This is not a criticism. It is a clarification. The consulting industry built enormous revenue on work that was valuable primarily because it was slow and expensive to do otherwise. That moat is gone.

That requires judgment. Not data. Not frameworks. Judgment built from having made similar decisions yourself, and from having been wrong enough times to recognize what wrong looks like before it announces itself.

This is why I&M was built the way it was. Every practitioner in this firm has operated at the level they advise on. We have run HR functions, structured deals, entered markets. We use AI aggressively. It makes our work faster and our analysis sharper. But the irreplaceable part of what we do is the judgment that directs it.

What this means for the companies we work with

If you are hiring a consulting firm today and they are not using AI in their delivery, they are slower and more expensive than they need to be. That is a problem.

If AI is the main thing they are selling you, ask one question: what is the judgment behind the model? Who is directing it, and what have they actually done?

The best consulting in 2026 is practitioner judgment, AI-accelerated. That is not a positioning statement. It is the only version that makes sense.