The Complete Mission Flow: Who Does What When
"I don't know what to do next."
You're joining a team where AI does 95% of the heavy lifting and you supervise, deploy, and verify. But if you don't know WHO to ask when stuck, WHEN to escalate vs. continue, and WHAT a good handoff looks like...
...you'll waste hours in confusion instead of shipping fast.
This guide shows the complete flow from "job posted on Upwork" to "client paid, testimonial received."
The Big Picture: 5 Phases
From job post to payment in 1-2 weeks
Emma
Inna
Rafael
Sofia
MayaThe AI Citizens Who Guide Each Mission
Hover over each citizen to see their role and philosophy.
Emma
The Scout
Filters noise, drafts winning proposals, and nurtures leads through systematic follow-up.
"I mirror the client's pain in their own words. No hype, just clarity and proof."
Inna
The Specifier
Writes complete 6-level documentation and locks scope via executable AC.md before code is written.
"Clarity emerges through structure, not accumulation. One solution per problem."
Rafael
The Guide
Generates 100% complete implementations from specs and provides DevOps support when developers get stuck.
"If it's not tested, it's not built. Code that compiles but hasn't run is theoretical."
Sofia
The Checker
Verifies DoD checklist, runs acceptance tests, and catches bugs before clients see them.
"Quality is non-negotiable. I verify every criterion, test every threshold, catch every bug."
Maya
The Bridge
Onboards clients, sends proactive updates, and coordinates Evidence Sprint demos and AC Green handoffs.
"I translate technical progress into business language. Clients always know where we are."
Strategic Support Throughout:
Alexis
The Strategist
Guides pricing decisions, tracks finances, and presents options for high-level strategic choices.
"Every proposal needs a price. Every month needs a dashboard. Strategy requires numbers."
Phase Details
ACQUIRE
~2 minEmma (AI)Find job on Upwork and draft winning proposal
SPECIFY
TBDInna (AI)Lock scope with complete 6-level documentation
BUILD
2-5 daysRafael (AI)Generate code, test locally, deploy to production
TEST
2-4 hoursSofia (AI)QA testing against AC.md criteria
DELIVER
~1 dayMaya (AI)Present demo, client accepts, payment received
Key insight: Each phase has a clear owner, AI support, and handoff criteria.
Phase 1: ACQUIRE (~2 minutes)
Search & Filter (Bigbosexf)
- Open Upwork, search for jobs matching criteria
- Copy entire results page
- Paste to Emma:
"Emma, analyze these jobs"
Emma's Analysis (AI, instant)
- Emma reads all job posts
- Returns:
"3 NO-GO (reasons), 2 STRONG GO (confidence scores)" - Provides brief rationale for each decision
Deep Dive (Bigbosexf)
- Open each STRONG GO job
- Copy ENTIRE job description (don't summarize!)
- Paste to Emma:
"Emma, write proposal for this job"
Emma Generates Complete Proposal (AI, ~60 seconds)
Emma produces:
- ✅ Full proposal text (plain text, ready to paste)
- ✅ Answers to client questions
- ✅ Metadata JSON (budget, timeline, client info)
- ✅ Telegram notification to team
Send (Bigbosexf)
- Review Emma's proposal (30 seconds)
- Copy-paste into Upwork
- Submit!
⏱️ Duration: ~2 minutes per proposal
10 minutes = 2 proposals sent + 8 minutes for coffee ☕
When to Escalate
- ❌ DON'T escalate: Emma says NO-GO → Trust her, move on
- ✅ DO escalate: Client responds with technical question you can't answer → Telegram NLR
Phase 2: SPECIFY (Duration: TBD)
[PLACEHOLDER - Multi-step process being documented]
This phase involves:
- Client onboarding (with Maya's guidance)
- Requirements gathering
- Writing complete specifications with Inna (6 levels: PATTERN → BEHAVIOR_SPEC → VALIDATION → MECHANISM → ALGORITHM → GUIDE)
- Locking scope via AC.md baseline
- Creating DoD (Definition of Done) checklist
What Gets Produced
- ✅ AC.md (Acceptance Criteria)
- ✅ DoD checklist
- ✅ Architecture notes
- ✅ Deployment guide
- ✅ Test plan
Phase 3: BUILD (2-5 days)
Read Specs (Kara, 30 min)
- Read Reanance's complete specs
- Check AC.md (what must work)
- Check DoD checklist (how we verify)
- Review architecture notes
- Check deployment guide
Generate Code (Rafael, AI, instant)
Kara asks Rafael:
"Rafael, generate implementation for Mission #47 per Reanance's specs"Rafael returns:
- ✅ Complete file structure
- ✅ All code files
- ✅ Dependencies (package.json, requirements.txt)
- ✅ Configuration (.env.example)
- ✅ Deployment instructions
Review & Test Locally (Kara, 2-4 hours)
- Create project from Rafael's code
- Install dependencies
- Run locally
- Test against AC.md criteria
- Fix any obvious issues
Deploy (Kara, 1-2 hours)
- Deploy to Vercel (frontend) or Render (backend)
- Configure environment variables
- Test deployed version
- Verify all AC.md criteria pass on production
Update SYNC.md (Kara)
## 2025-11-05 14:30 - Kara: Mission #47 Deployed **Status:** Deployed to production, ready for QA **URL:** https://client-project.vercel.app **What works:** All 5 AC.md criteria tested and passing **Next:** Bigbosexf QA testing
⏱️ Duration
- Simple (landing page, bot): 2 days
- Medium (CRUD dashboard, API integration): 3-4 days
- Complex (multi-feature app): 5 days
When to Escalate
If stuck >1 hour:
- Try asking Rafael first:
"Rafael, I'm getting this error..." - If still stuck after Rafael's guidance → Telegram NLR
- If truly urgent/blocking → Call NLR
Phase 4: TEST (2-4 hours)
Bug Fix Loop
When to Escalate
- ❌ DON'T escalate: Bug found → Talk to Kara directly
- ✅ DO escalate: Bug is critical AND Kara can't fix in 2 hours → Telegram NLR
Phase 5: DELIVER (~1 day)
[PLACEHOLDER: Payment flow details TBD]
[PLACEHOLDER: Team payment distribution TBD]
"Who Do I Ask?"
| Situation | Ask | Why |
|---|---|---|
| Code error <1hr | Rafael (AI) | He debugs instantly |
| Code error >1hr | NLR (Telegram) | Need human guidance |
| Bug in QA | Kara (direct) | She's the implementer |
| Client question | Reanance (with Maya/Rafael) | Reanance owns client relationship |
| Scope change | Reanance decides (or NLR) | Reanance owns scope decisions |
| Job evaluation | Emma (AI) | She evaluates fit |
| Ready for delivery? | Sofia (AI) | She verifies DoD |
Escalation Rule
Stuck >1 hour? → Try AI first → Still stuck? → NLR
Urgent/blocking? → Call NLR
Quiz: Test Your Understanding
Scenario 1: You're Kara. You deployed the app but get a weird error you've never seen. You've been stuck for 45 minutes. What do you do?
Scenario 2: You're Bigbosexf doing QA. You find a bug where the contact form doesn't send emails. What do you do?
Scenario 3: You're Reanance. Mid-implementation, the client says "Can we also add a blog section?" What do you do?
Quick Reference Card (Print & Keep)
STUCK? WHO TO ASK
- Code error <1hr → Rafael (AI)
- Code error >1hr → NLR (Telegram)
- Bug in QA → Kara (direct)
- Client question → Reanance (with Maya/Rafael)
- Scope change → Reanance decides (or NLR)
- Job evaluation → Emma (AI)
- Pricing guidance → Alexis (AI)
- Ready for delivery? → Sofia (AI)
ESCALATION RULE
Stuck >1 hour? → Try AI first → Still stuck? → NLR
Urgent/blocking? → Call NLR
Next Steps
Now you understand WHO does WHAT WHEN. Next articles cover:
- How to Talk to AI Citizens - Get 3-5x faster by asking good questions
- What Good Documentation Looks Like - So Rafael generates correct code first time
- Testing Mindset: AC Green or It Didn't Happen - How Sofia checks your work
Ready to start? Ask NLR to assign you your first shadow mission (watch someone else's mission end-to-end).