From an ambiguous brief to a decision, shown.
Stakeholders rarely arrive with a clean question; they arrive with a problem and, often, a solution already in mind. My job is to run the engagement that tests it: reframe the real question, direct a team and an AI-augmented research method against it, and come back with a decision, their path validated or a better one found.
Most analysts use AI to produce faster. I use it to think wider, pressure-test harder, and decide sharper.
Prompts shown are representative patterns from my working method, generalized and anonymized; the actual workflow has complex chained prompts, with constraints and several other controls built in.
Stage 1 of 5
Scope and frame the mandate
I run point with stakeholders directly. They bring a problem, often with an answer already in mind; I reframe it into the real, answerable question and pressure the assumption that their preferred solution is the right one.
I use AI to map the problem space and surface alternative framings and hidden assumptions before anything gets scoped.
Here is the brief exactly as the stakeholder gave it: [brief]. Separate what is being asked from what is being assumed. List every assumption the framing smuggles in, rank them by how much the answer changes if they are wrong, and propose three alternative framings of the underlying question, each with the evidence that would make it the right one to pursue.
A scoped mandate leadership signs off on, and clarity on whether we are validating their hypothesis or finding a better one.
Run this way, a leadership GTM hypothesis became defined target-market priorities for a launch, and the engagement grew 50% over two years.
Stage 2 of 5
Architect the research
I design the approach, the hypotheses, sources, and analytical model, and set the standard my team of five analysts delivers to.
AI drafts competing hypotheses and source maps, so coverage gaps surface on day one instead of in the final review.
We are testing this hypothesis: [hypothesis]. Draft the three strongest competing hypotheses a skeptical reviewer would raise. For each, specify the evidence that would confirm or kill it, the most credible source types for that evidence, and where our coverage is likely to be thinnest. Output it as a source map a team of analysts can execute against.
A research plan and a quality bar the team executes against.
This is the standard a team of five analysts delivers to, set once and enforced through the engagement.
Stage 3 of 5
Direct retrieval and synthesis
I direct the team and our AI-enabled workflows to gather and synthesize evidence at scale across primary and secondary sources, staying the quality gate and upskilling analysts as we go.
AI accelerates first-pass retrieval and synthesis across large, messy corpora, turning a pile of documents into structured evidence the team can interrogate.
Across these documents on [market]: extract every quantitative claim relevant to [question] into a table of claim, value, source, date, and the method behind the number where stated. Flag conflicts between sources instead of averaging them, and close with a list of what this corpus cannot answer.
A synthesized evidence base, not a stack of links.
Workflows built on this pattern cut per-report turnaround by 20% and freed analyst time for the higher-judgment work.
Stage 4 of 5
Pressure-test the thinking
I personally red-team the findings, and the stakeholder's original solution, against counter-evidence and edge cases, so what reaches the client survives scrutiny.
I use AI as an adversary, generating the strongest counter-argument and flagging where we are over-reaching.
Act as the strongest critic of this recommendation: [draft]. Build the best counter-argument the evidence allows, identify where we are extrapolating beyond the data, and list the three questions a hostile executive would ask that we cannot yet answer.
A view that holds up, whether it confirms the client's instinct or challenges it.
Nothing reaches the client until it has survived the strongest counter-argument we could build against it.
Stage 5 of 5
Advise the decision
I translate the analysis into a clear recommendation, validate the path or put a research-backed alternative on the table, with the rationale, risks, and implication for roadmap, GTM, or investment.
AI helps shape tight narratives and decision-ready artifacts, so the thinking lands in the room instead of in an appendix.
Here is the full analysis: [synthesis]. Compress it into the decision structure: the recommendation in one sentence, the three strongest reasons with their evidence, the main risk and its mitigation, and what changes for roadmap, GTM, or investment if leadership says yes.
A decision the client can act on, and a relationship that grows into the next mandate.
Years of bi-monthly readouts to senior leadership, and the first call when the market moved.