Ship to Prod AI Hackathon · San Francisco · April 24, 2026
Every enterprise with real capability — every machine shop, fleet, or lab — wastes two weeks integrating with an agent marketplace before they earn a dollar. Navi compresses that to one phone call.
docs/media/navi-demo-2min.mp4 · 114 seconds · 1920×1080 · 4.1 MB
| Channel | URL / Contact |
|---|---|
| 📞 Voice agent | +1 (650) 448-0770 |
| 💬 Chat console | https://pcc-operator-backend-production.up.railway.app/ |
| 🤖 Operator dashboard | https://pcc-operator-backend-production.up.railway.app/op?id=<session> |
| 🧠 Guild-published agent | https://app.guild.ai/agents/globalmysterysnailrevolution/pcc-enterprise-onboarder |
| ❤️ Health | https://pcc-operator-backend-production.up.railway.app/health |
| Sponsor | Status | Role |
|---|---|---|
| Vapi | ✅ LIVE | voice onboarding (phone number live) |
| Guild | ✅ LIVE | published llmAgent v1.0.1, public |
| Nexla | ✅ LIVE | real data pipelines (sources 120398+ in dataops.nexla.io) |
| InsForge | ✅ LIVE | Postgres backend auto-provisioned per enterprise |
| Redis | ✅ LIVE | RedisVL capability index + XADD a2a intent stream |
| Coinbase CDP | ✅ LIVE | viem-generated Base Sepolia wallet per operator |
| TinyFish | ✅ LIVE | SSE agent scrapes enterprise websites |
| Senso (cited.md) | ✅ LIVE | content engine draft published (content_id tracked) |
| Chainguard | ✅ LIVE | cgr.dev/chainguard/node base image on Railway |
| agentic.market | 🟡 POST attempted, graceful fallback | |
| x402 | 🟡 middleware scaffolded, Coinbase facilitator wired |
8 fully real, 2 partial — see full breakdown in ARCHITECTURE.md
The inevitable judge question is "three databases, really?" Here's the honest answer:
📚 InsForge ← the operator's books (durable, ACID, one per enterprise)
📓 Ghost ← the verifier's scratchpad (forked Postgres per capture, discarded after)
⚡ Redis ← the agent's nervous system (µs working memory + a2a Streams bus)
Same data shape, three different latency × durability points. Full sequence diagram in ARCHITECTURE.md.
┌─ phone / chat / Guild ────────────┐
│ ↓ │
│ Navi backend (Express on │
│ Chainguard image, Railway) │
│ ↓ │
│ TinyFish Nexla (discovery) │
│ ↓ │
│ InsForge Redis Ghost (storage)│
│ ↓ │
│ CDP → x402 → agentic.market │
│ Senso → cited.md │
│ Guild deploy │
│ ↓ │
│ 🤖 new operator agent ONLINE │
└───────────────────────────────────┘
See docs/ARCHITECTURE.md for 6 Mermaid flowcharts with full detail.
navi/
├── packages/backend/ ← Express + 8 real sponsor wrappers + Dockerfile (Chainguard)
├── packages/guild-agent/ ← published Guild llmAgent source
├── apps/voice/ ← Vapi assistant config + Task Runner prompt
├── docs/
│ ├── media/navi-demo-2min.mp4 ← 2-min demo video
│ ├── media/voiceover.mp3 ← raw voiceover (AndrewNeural, 114s)
│ ├── ARCHITECTURE.md ← 6 Mermaid diagrams
│ ├── DEMO-3MIN.md ← stage script (3-min)
│ ├── VIDEO-2MIN.md ← video storyboard
│ ├── SUBMISSIONS-FINAL.md ← per-track copy for Devpost
│ └── PLAN-V3.md
└── fixtures/oakland-titanium-mills/ ← demo enterprise
git clone https://github.com/LamaSu/navi.git
cd navi/packages/backend
cp .env.example .env.local # fill in keys (see docs/INTEGRATION-TASKS.md)
pnpm install
pnpm dev # → http://localhost:3000Or hit the live Railway deploy directly — see the URLs above.
Built in ~6 hours at Ship to Prod · AWS Builder Loft SF · April 24 2026