The decision you are actually making

The decision is whether the process stays and the tool is bolted on, or whether the process is redesigned around it.

The first is why adoption stalls: the same work, marginally faster, with a subscription attached. The second is harder, slower to start, and the only version that changes a number on the P&L before the hold period ends.

What I do

An AI maturity diagnostic spanning process-level automation through full agentic redesign, by process rather than by company
A prioritised roadmap sequenced by what compounds, against the clock you are actually running
Hands-on build support: I build these systems, not only advise on them

What you get

An honest read of where adoption actually stands
The build, where you want it built rather than described
A view on what to buy, what to build, and what to leave alone

If the company is Singapore-based, part of the build is fundable: Singapore public money for SME AI adoption, mapped in the 2026 grants handbook.

The record

Every system below is one I designed or operate. None of them is a private equity portfolio company. They are the proof I can show you: a corporate investing standard, a growth unit's discovery pipeline, and the estate this practice runs on. The portfolio-company version is the same work under a hold-period clock.

AI-powered investment screening, adopted as the client standard, 2025

A leading Asian firm where corporate investing, M&A and the core business each assessed targets their own way: three groups, three vocabularies, three spreadsheets, no shared answer. I designed the investment thesis and sourcing framework first, because a screening engine without a thesis is a faster way to be busy. Then I built the evaluation workflow on top. A large company universe screened to a qualified shortlist, every target scored the same way, each carrying a recommended route: partner, invest, or acquire. Read the case.

Innovation discovery on AI rails, 2025

The non-core growth unit of a global consumer group asked whether its slowest, most expensive stages could run materially faster without losing rigour. I moved the entire discovery front end onto AI rails: clustering thousands of market signals into themes, generating candidates inside each theme rather than waiting for a workshop, then filtering and ranking. The validation was not a survey. People paid with their own money in a live trial. The same sweep threw off a pipeline of real companies for the client's investing and M&A teams. Read the case.

The operating estate I run this practice on, 2025 to present

Several AI sessions working in parallel, a coordinator handing tasks to builder agents, and nothing shipping until an independent check has tried to break it. I did not buy this. I built it, I operate it daily, and it is the clearest proof I can offer of the capability I sell. Read the case.

A portfolio-company build takes the same shape.

A B2B sales engine at SME scale, 2025 to present

Enterprise sales discipline: pipeline stages, clocks on every deal, a system that remembers every conversation, rebuilt so a lean team or a single operator can actually run it. Same standard, without the team it usually takes. Read the case.

What the first call looks like

Send the materials behind the decision ahead of time. I come to the 30-minute conversation having read them, with a written first view: three things I can already see, and the one question a first phase would have to answer. Where I am the wrong person for it, I say so on the call rather than after it. I run the work and I answer for the output.

A fit turns into a diagnostic sprint next: two to three weeks, one question, closing on a one-page verdict and a go or no-go on anything further. How an engagement actually runs.

Start with the process, not the tool

A short conversation about the process you would redesign first.

Send a message