Skip to main content

How does value flow today — and where does time die?

Current State: RFP Processing (Form-Based)

VALUE STREAM MAP — RFP RESPONSE (CURRENT)
════════════════════════════════════════════════════════════

[RFP PDF ARRIVES]


┌──────────────────┐
│ 1. Download PDF │ C/T: 1 min Wait: 0 %C&A: 100%
└────┬─────────────┘
│ ░░░░░░░░░░░░░░░ (find the right module: 5 min)

┌──────────────────┐
│ 2. Open Venture │ C/T: 2 min Wait: 5 min %C&A: 90%
│ Create Form │
└────┬─────────────┘
│ ░░░░░░░░░░░░░░░░░░░░░ (copy-paste questions: 30 min)

┌──────────────────┐
│ 3. Paste Questions │ C/T: 30 min Wait: 0 %C&A: 70%
│ One By One │ (miss appendix questions)
└────┬─────────────┘
│ ░░░░░ (trigger AI: 1 min)

┌──────────────────┐
│ 4. Run AI Loop │ C/T: 3 min Wait: 1 min %C&A: 80%
└────┬─────────────┘
│ ░░░░░░░░░░░░░░░░░░░░░░░░░ (review answers: 60 min)

┌──────────────────┐
│ 5. Review + Edit │ C/T: 60 min Wait: 0 %C&A: 85%
└────┬─────────────┘


[PROPOSAL DRAFT]

Total Cycle Time: 96 min
Total Wait Time: 6 min
Total Lead Time: 102 min
Flow Efficiency: 94% (wait is low — the WORK is the waste)

════════════════════════════════════════════════════════════

Future State: RFP Processing (Conversational)

VALUE STREAM MAP — RFP RESPONSE (CONVERSATIONAL)
════════════════════════════════════════════════════════════

[RFP PDF ARRIVES]


┌──────────────────┐
│ 1. Drop PDF in │ C/T: 10 sec Wait: 0 %C&A: 100%
│ Chat │
└────┬─────────────┘


┌──────────────────┐
│ 2. Agent Extracts │ C/T: 30 sec Wait: 0 %C&A: 95%
│ All Questions │ (vision model, no copy-paste)
└────┬─────────────┘


┌──────────────────┐
│ 3. Agent Runs │ C/T: 2 min Wait: 0 %C&A: 80%
│ RFP WorkChart │ (streams progress)
└────┬─────────────┘


┌──────────────────┐
│ 4. Human Reviews │ C/T: 30 min Wait: 0 %C&A: 90%
│ in Chat Thread │ (inline editing)
└────┬─────────────┘


[PROPOSAL DRAFT]

Total Cycle Time: 33 min
Total Wait Time: 0 min
Total Lead Time: 33 min
Flow Efficiency: 100%

════════════════════════════════════════════════════════════

Comparison

MetricCurrent (Forms)Future (Conversation)Improvement
Lead time102 min33 min68% reduction
Copy-paste step30 min manual30 sec automated98% reduction
Questions missed~30% (appendix)~5% (vision model)83% fewer misses
Context switches4 modules1 chat window75% fewer switches
Navigation time5 min0 minEliminated

Seven Wastes Applied

WasteCurrent StateConversational Fix
MotionNavigate 4 modules per taskSingle chat window
TransportCopy-paste PDF to formDrop file, agent extracts
ReworkMissed questions require re-extractionVision model catches appendices
Over-processingManual formatting of extracted textAgent preserves structure
WaitingFind correct module (5 min)Agent routes automatically
InventoryMultiple tabs open simultaneouslyOne conversation thread

Headline

96 minutes to 33. One window.

P-Q-D

ProblemQuestionDecision
30 min copy-paste step in a 96-min processWhat if the document came to the agent?PDF drop + vision extraction eliminates the bottleneck

Gate

  • All steps documented with cycle time
  • Wait time measured between steps
  • Flow efficiency calculated (94% current, 100% future)
  • Bottleneck identified (#3 copy-paste: 30 min)
  • Waste types categorised (motion, transport, rework primary)
  • Improvement opportunities prioritised (PDF extraction > voice > proactive agent)

Context