Role, Corporate Development

Claude for
corp dev

The pipeline machinery around M and A: sourcing, triage, diligence synthesis, memo drafting. The work where a better first draft saves days, not hours.

This is my actual day job, so I've spent more time than anyone should thinking about where Claude fits in the corp dev workflow. The short version: anywhere that involves turning a pile of unstructured information into a structured decision. Long-lists, data rooms, stakeholder inputs, memo writing. Every one of these is a perfect fit for the four-elements pattern from the Beginner guide.


Three places to start

Three workflows that would each justify the tool on their own, and which together change the rhythm of the job.

1. Triaging a sourcing long-list

You have 150 names, a set of criteria, and a deadline. Paste the long-list, paste the criteria, and ask Claude to rank and flag. The output is a ranked shortlist with reasons tied to your actual criteria, plus a "need more info" list of companies where the data was too thin to judge. This is the most time-saving single prompt in the whole corp dev workflow.

Triage prompt, annotated

Four elements applied to the single highest-leverage corp dev prompt.

YouHere is a long-list of 150 SaaS companies in the logistics tech space. 1 Our investment criteria, pasted below: - revenue between £5m and £30m - EBITDA positive in the last 12 months - UK or EU headquartered - not owned by a strategic 2 Rank the top 20 against these criteria. For each, one line on why it made the cut. 3 For any company where the data is too thin to judge confidently, move it to a separate "need more info" list with the specific question that would unlock a decision. Do not score criteria you have no data for. 4
1
Context. Theme, sector, geography. Claude needs the frame.
2
Example. The criteria themselves. Worth pasting in full rather than describing.
3
The ask. Ranked top 20 with one-line rationales. Specific output shape.
4
Escape hatch. Thin-data companies go into their own bucket with a specific unlock question. No score-by-guess.

Where this falls over

Claude will happily score a company on criteria where it has no data, unless you explicitly tell it to move thin-data companies to a separate list. Always include that instruction.

2. Synthesising a data room into a memo draft

A data room is a few hundred documents you need to turn into a ten-page memo. Upload the key docs to a Project, paste your last three memos as style references, and ask Claude for a first draft. You will edit it heavily, but you will be editing rather than writing from scratch. Two or three days of work becomes one.

Asking for a memo vs actually getting one you can use

The difference between confident fabrication and a draft grounded in the source.

Youwrite me an investment memo for Orinoco Logistics
What Claude sends back

A plausible-sounding memo with a neat executive summary, confident revenue figures that do not appear in any document, a "growth trajectory" paragraph that reads well, and a clean risk section. Looks good until you realise half of it is fabricated.

YouI am drafting an investment memo for Orinoco Logistics. Attached: the data room index and the six key documents I have pulled out. Attached: our last three investment memos, as style and structure references. Produce a first draft matching the structure of our previous memos. Focus particularly on operational risk and concentration of revenue. Do not invent any number. Every figure must come from the source docs, with the document and page cited inline. List every inconsistency you find across the sources. Say "not in the source" for anything you cannot find rather than filling it in.
What Claude sends back

A first-draft memo matching your previous structure, every number cited back to a specific source document.

Inconsistencies to resolve before this goes anywhere:
(1) The financial summary in Q3 board minutes shows EBITDA of £4.2m. The management accounts appendix shows £3.8m for the same period.
(2) The data room index lists "customer concentration analysis" as document 14. I could not find document 14 in the files I was given.
(3) The narrative memo from the founder mentions "three top customers representing 45% of revenue". Your last three memos treat anything above 35% as a red flag. Should this be in the risk section?

Where this falls over

The first draft will invent numbers if you don't tell it not to. The prompt must say: do not invent numbers, flag every inconsistency you find across the source documents, say so if something is missing.

3. Pipeline management and status updates

The weekly pipeline review involves pulling the state of twenty live opportunities into a readable update for leadership. Give Claude the CRM export plus last week's update, and it will produce this week's version in the same voice and format. You spend the saved time on the conversations the update triggers.

TR
Tom R. #corp-dev Monday, 09:04

Stopped writing Monday pipeline updates from scratch. Now I paste last week's update into Claude along with this week's CRM export and tell it "match the voice of last week, flag any deals that have gone stale since then, don't invent progress that isn't in the CRM".

A 45-minute job has become about ten. Leadership has not noticed any drop in quality. That is the whole point.

11 clap 6 fire

Where this falls over

If the CRM data is messy (inconsistent stages, outdated notes), Claude will reflect that messiness. The fix isn't Claude, it's the data. Clean the CRM first, then automate.


Worked examples

Before and after, on tasks from corp dev work.


Prompts to steal

Copy any of these, fill in the square brackets, run.


Try this week

Pick one. All of these are under an hour.


The bit of corp dev that Claude cannot do is the judgement call. It can triage, synthesise, draft, and reformat, but it cannot tell you whether a deal is the right deal. That part stays with you. Everything else is on the table.

If you want to go broader, the Beginner guide covers the habits that apply everywhere. The prompt library has the full set of role-tagged prompts in one place.