Bid response checklist: what AI handles, what humans still own
Two questions sort every section of a tender in about thirty seconds. Is the answer already in your sources, and who takes the call if it's wrong?
Read post
Field notes
Proposals, SOWs, compliance evidence, RFP responses, and what we're learning as we build SkyDraft, what works inside professional services firms, and what doesn't. New posts every Monday, Wednesday, and Friday.
Two questions sort every section of a tender in about thirty seconds. Is the answer already in your sources, and who takes the call if it's wrong?
Read post
Three people use it daily, eleven use it occasionally, and the rest went back to copying last quarter's file. Nobody cancelled anything. Nothing was ever set up.
Read post
Ninety seconds for 1,600 words of clean prose, then an afternoon fixing structure, facts, and voice. The model was never the limitation.
Read post
Your firm's confidentiality position is currently being set at 4pm on a Friday by whoever is under the most pressure. Eight questions that fix that.
Read post
You own the best document archive in your industry and you still start from a copy. The archive holds the output of your standard, not the standard itself.
Read post
Eight paragraphs of clean consulting prose, and not one sentence she could defend in the closeout meeting. The prose was never the problem.
Read post
A fourteen-day tender where the first complete draft lands on day eleven gives you three days to win it. The number worth moving isn't writing speed.
Read post
Ask a partner what a good report looks like and you get a confident twenty-minute answer. That answer is a firm asset stored in the least available place you own.
Read post
Three demos in a week and you can't tell them apart, because every demo is built around the one thing that's easy. Stop being the audience and run these tests yourself.
Read post
A hallucinating tool and a grounded one look identical on a good day. The difference shows up on the facts you didn't provide. Five probes to find out which you have.
Read post
Voice is how you write. Identity is who's writing. Get the tone perfect and the draft still argues from nowhere. Here's the asset most firms skip.
Read post
Autopilot gives you fourteen pages you have to audit against the source. A review gate on every section is the only 'faster' that ships.
Read post
Requirements, deliverables, features: the list sections are where generic AI invents, drops, and flattens. Draft them one item at a time.
Read post
One-shot optimises the demo. Section-by-section optimises the deliverable. The same 22-page proposal, run both ways.
Read post
Auditors don't reward fluency. They reward precision. Why generic AI drafts audit-failing narratives and what a structured workflow does instead.
Read post
Automate the 60% that's recycled. Keep the 40% that decides the win human. A section-by-section breakdown for bid managers.
Read post
Voice is the product. Generic AI flattens firm voice into corporate sameness. Here's what a well-designed AI workflow does instead.
Read post
Minute-by-minute worked example: 90-min transcript to 20-page assessment report in twenty minutes, with the AI asking the right questions along the way.
Read post
Hallucination is a design choice. The alternative is a workflow that lets the AI ask questions instead of guessing plausibly.
Read post
Same chassis, different document. Why generic AI drafts proposal-flavoured SOWs, and how to set up a workflow that gets the contractual specificity right.
Read post
Six patterns that hold up under a real bid deadline, and four that feel productive but cost you the win.
Read post
The thinking is yours. The typing is the bottleneck. Here's how to remove it without losing your firm's voice.
Read post