Frank advice, with the working shown.

Candor is led by Izam Ryan MBA FCA, a chartered accountant who builds with AI and measures what it changes. He trained at PwC and spent ten years across transactions and advisory; most recently, as an employee of an advisory firm, he led eleven financial and vendor due diligence engagements for private equity and trade buyers, mostly software businesses. He directs frontier models to build working tools, and built the harness that makes their analysis re-performable. Thought Leader, The AI Journal, with four essays published April to May 2026.

There was no single moment. AI came into my diligence work as a trickle, then a flood, then an avalanche: first a summary, then a schedule, then whole populations of transactions analysed where a team would once have sampled. The more it did, the clearer the gap became. The analysis was there. What was missing was everything that lets me put my name to it: guardrails against a model inventing a figure, a way to manage the risk of relying on it, and above all the ability for someone who was not there to re-perform the work and get the same answer. That gap is what I work on now, and it is the same gap in every business spending on AI: the gap between what AI was meant to return and what shows up in the numbers, with nobody able to show the working behind either. Every figure I produce traces to the document it came from, and the working stays beside it.