GoogleCloudPlatform Acquisition Target Screening Playbook
Methodology for the "recommend companies that might be good targets" mode.
Inputs to elicit or infer
- Acquirer profile / thesis: therapeutic areas of interest, modality preferences (small molecule, biologics, ADC, cell/gene therapy, RNA), stage appetite (platform vs. de-risked late-stage vs. commercial), approximate deal-size budget, and strategic gaps to fill.
- If the user gives only a vague ask, state the assumptions you screen under.
Screening funnel
- Generate candidate universe — use Google Search for recent pipeline news and analyst M&A speculation, and EDGAR full-text search to find companies whose filings discuss the target technology/indication.
- First-pass filter — modality/indication fit, stage fit, and rough size fit. Discard obvious mismatches; keep 8–12 names.
- Verify corporate independence & M&A status (CRITICAL - BATCH PARALLEL) —
for all 8–12 candidate survivors from Step 2, execute verification searches
in parallel within a single turn (do not query one-by-one serially):
- Query format:
"<Company>" (acquired OR acquisition OR buyout OR merger OR "deal closed" OR delisted)or"<Company>" acquired acquisition merger buyout - For public companies, verify active trading and actively filing (check for Form 15/25 delisting notices or 8-K/6-K merger completions).
- Immediately discard all acquired/merged/delisted entities.
- Advance only verified independent standalone survivors to scoring.
- Query format:
- Score survivors — for each, a lightweight version of the five diligence pillars. Emphasize: strategic fit, catalyst timing, and (for public names) cash runway as a negotiating-leverage / urgency signal.
- Rank — produce a ranked shortlist with a fit score and a one-line thesis per name.
Output format (screening mode)
A ranked table:
| Rank | Company | Ticker | Lead asset / platform | Stage | Fit rationale | Key risk | Est. cash runway | Source |
|---|
Follow with 2–4 sentences per top candidate expanding the thesis, and a note on what deeper diligence (a full whitepaper) would resolve next.
Guardrails
- Target Acquirability Rule: Every recommended target MUST be an independent, standalone entity. Never recommend companies that have already been acquired, merged into another entity, or are operating subsidiaries of larger biopharma.
- No Assumed/Fabricated Filings: Never invent or extrapolate quarterly filing periods (e.g. unretrieved current-year 10-Q/10-K filings) for an acquired or delisted company. If a company has ceased filing or has no recent filings in the current calendar year, investigate whether it was acquired or taken private.
- Only recommend names you can support with at least one concrete source.
- Distinguish public (screenable via EDGAR) from private (news) candidates; flag data limitations for private ones.
- Never present speculation as a confirmed deal rumor; attribute rumors.
- Offer to produce a full whitepaper on any shortlisted name.