Do the project work
28 primitives in this stage — 25 skills · 3 agents. Concept-stage catalogue, kept vendor-agnostic.
Structure raw interview capture →
"I took twelve pages of notes and couldn't find the insight I needed two days later — now I process transcripts immediately while they're still fresh."
First-pass relevance triage →
"The client sent a SharePoint folder with 200 files; I spent two days reading things that didn't matter — there had to be a faster way."
Red-team challenge loop →
"I want to hear the hard questions before the client does, not in the room — the red-team loop is like hiring a sceptic for an hour."
Draft the evidence collection plan →
"I always start a project meaning to gather 'whatever we need' — this forces me to decide what that actually is before a single interview is booked."
Stress-test plan for coverage and bias →
"We submitted findings and the client immediately asked 'but did you talk to operations?' — now I run this before we set foot in the field."
Generate a tailored interview guide →
"Generic interview guides get generic answers — I need questions shaped around what this person actually knows."
Design the survey or observation instrument →
"I've launched a survey and got data I couldn't use because the questions were subtly wrong — now I design the instrument before I build it."
Assess evidence sufficiency and prioritise gaps →
"A coverage diff tells you what's missing; this tells you whether it matters enough to delay the analysis or whether you can qualify the finding."
Issue tree / hypothesis tree builder →
"Without a proper issue tree I'm just doing random analysis and hoping it adds up to something — the tree forces me to define what 'solved' actually means."
Key variable & relationship mapper →
"Every engagement has a moment where you realise the variables you've been measuring aren't the ones actually driving the outcome — this surfaces it early."
Analytical approach designer →
"Choosing the wrong analytical approach wastes a week and produces conclusions the client can't trust — designing the approach before running the numbers prevents that."
Cross-output synthesiser →
"The interview findings said one thing and the model said another — the synthesiser helped me see that they were actually both right, about different things."
'So what' insight extractor →
"I can describe what the data shows all day; this is the thing that forces me to say what the client should actually do differently because of it."
Narrative arc drafter →
"I had all the right insights but couldn't land them in the room — the narrative arc makes the logic unavoidable rather than just defensible."
Recommendation prioritiser →
"Clients can't act on fifteen recommendations of equal weight — this collapses them into a sequence the client can actually execute."
Evidence traceability checker →
"A senior partner once asked me to 'show the evidence' for a claim I thought was obvious — I couldn't, and the whole section fell apart in front of the client."
Shape the solution's structure and design logic →
"The recommendations were right but the solution design was vague — every implementation question exposed a gap we hadn't thought through at design time."
Stress-test the proposed design against requirements and constraints →
"We built a solution the client loved until they tried to implement it — pressure-testing the design would have caught the constraint we'd ignored."
Draft a section of the core deliverable against the design →
"Staring at a blank slide with five minutes before the team review is the worst feeling — a drafted section I can react to is always faster than starting from nothing."
Run an internal quality review of the built solution →
"We sent out a deliverable with an inconsistency on slide 14 that the client spotted before we did — now internal QA is non-negotiable before anything leaves the team."
Probe the solution against edge cases and failure conditions →
"Our solution worked beautifully for the 80% case and fell apart on the first exception the client tried — probing edge cases is what separates a demo from a real solution."
Triage review feedback into a prioritised change plan →
"Fourteen comments from three reviewers and I didn't know where to start — triaging first meant I fixed the structural things before the wordsmithing."
Apply an agreed revision to the working draft →
"Applying feedback quickly and cleanly — without accidentally breaking the section before it — is a skill that separates junior from senior analysts."
Draft the status report from progress facts →
"Status updates I write from scratch always take too long and never say the same thing twice — a consistent format that just needs the facts filled in is much better."
Draft the change request →
"I'd done the extra work before writing up the change request and lost the leverage — now I draft the CR the moment I see the scope shifting."
Prepare the negotiation brief →
"Walking into a change order conversation without a brief means I give ground I didn't intend to — the negotiation brief keeps me anchored to what matters."
Draft the risk escalation with options →
"Clients don't want surprises — they want to be told early, with options, not handed a problem and no path forward."
Draft the revised delivery plan →
"We lost a week to scope creep and I needed a credible revised plan in the client's inbox by morning — drafting from scratch under pressure always produces something fragile."