Flow Diagrams
Four maps. Each answers the question that feeds the next. The flywheel only spins when the library has entries — everything starts with seeding.
What does success look like?
Three outcomes. Each proves a different layer of the thesis.
OUTCOME MAP — RFP Engine
═══════════════════════
┌────────────────────────────────────────┐
│ WIN RATE TRENDS > 30% │
│ (Better bids, less wasted effort) │
└──────────────────┬─────────────────────┘
│
┌──────────────┴──────────────┐
│ │
┌──────────▼──────────┐ ┌──────────▼──────────┐
│ AUTO-FILL > 50% │ │ LIBRARY 10+ ENTRIES │
│ │ │ WITHIN 14 DAYS │
│ Each bid makes │ │ │
│ the next faster │ │ Break chicken-and- │
│ │ │ egg with manual seed │
└─────────────────────┘ └──────────────────────┘Outcomes defined. Where does time actually die?
Where does time die?
20-40 hours per RFP. 70% is duplicate effort.
VALUE STREAM — Current State
════════════════════════════
RFP LANDS FIND OLD BIDS COPY-PASTE EDIT & SUBMIT
───────── ───────────── ────────── ─────────────
┌────────────┐ ┌────────────┐ ┌────────────┐
3 weeks ────▶ │ Search old │───▶│ Copy from │───▶│ Edit every │
deadline │ Word docs │ │ Word to new │ │ answer │
│ 4-8 hours │ │ 8-16 hours │ │ 8-16 hours │
└────────────┘ └────────────┘ └────────────┘
WASTE: 70% is finding + copying answers that exist somewhere
TOTAL: 20-40 hours per RFP
VALUE STREAM — Target State (with compound library)
═══════════════════════════════════════════════════
RFP LANDS AUTO-FILL SME REVIEW SUBMIT
───────── ───────── ────────── ──────
┌────────────┐ ┌────────────┐ ┌────────────┐
3 weeks ────▶ │ Library │───▶│ Approve or │───▶│ Generate │
deadline │ matches 70% │ │ edit 30% │ │ deck │
│ < 1 hour │ │ 4-8 hours │ │ < 1 hour │
└────────────┘ └────────────┘ └────────────┘
SAVED: 12-24 hours per RFP ($1,150 value per bid)
COMPOUND: Each bid adds answers. Next bid is faster.Time saved identified. What must happen first?
What must happen first?
The flywheel has a cold-start problem. Manual seed breaks the deadlock.
DEPENDENCY MAP
══════════════
CRM Contacts + Deals (LIVE)
│
▼
Venture CRUD (LIVE) ──────────────────────────────────┐
│ │
▼ ▼
Q&A Management (LIVE) Go/No-Go Scoring (GAP)
│ "Stop before you start"
▼
Answer Editor + AI Draft (LIVE/BUILT)
│
▼
┌─────────────────────────────────┐
│ MANUAL SEED (CRITICAL PATH) │
│ Import 5 past bids, 20+ Q&A │
│ This breaks chicken-and-egg │
└────────────────┬────────────────┘
│
┌───────────┴───────────┐
▼ ▼
SME Approval Queue Answer Library
(BUILT, empty) (BUILT, empty)
│ │
└───────────┬───────────┘
▼
Auto-fill Algorithm (LIVE)
│
▼
Analytics Dashboard (BROKEN)Dependencies mapped. How does the flywheel actually spin?
How does the flywheel spin?
One loop. Each bid adds answers. The library grows. The next bid is faster.
THE COMPOUND LOOP
═════════════════
┌──────────────┐ ┌──────────────┐ ┌──────────────┐
│ NEW VENTURE │────▶│ AUTO-FILL │────▶│ ANSWER │
│ (RFP lands) │ │ from library │ │ EDITOR │
└──────────────┘ └──────────────┘ └──────┬───────┘
▲ │
│ AI draft OR
│ manual write
│ │
│ ▼
┌──────┴───────┐ ┌──────────────┐ ┌──────────────┐
│ NEXT BID │◀────│ LIBRARY │◀────│ SME REVIEW │
│ is faster │ │ (grows each │ │ approve / │
│ │ │ bid cycle) │ │ reject │
└──────────────┘ └──────────────┘ └──────────────┘
Each rotation: library grows → fill rate increases → time saved grows
The loop IS the product. Without approved answers, it's a project board.Questions
What does the value stream map look like when scoped to RFP only — and is the bottleneck bid prep or library adoption?
- If the library stays empty after manual seeding, is the problem the tool or the workflow?
- At what library size does the auto-fill rate cross from "nice to have" to "can't work without it"?