Editorial foundation
LockedCMS, creator review, membership, and multilingual governance are specified.
INVESTOR READINESS · PUBLIC EDITION
Truth infrastructure for how a city is understood.
Tram Saigon is developing from a trusted editorial system into a City Intelligence Engine: sourced content, governed AI, accountable creators, and verifiable city data.
This is a strategy and readiness dashboard. It is not a securities offering, revenue forecast, or valuation commitment.
EVIDENCE STATUS
Current truth
We separate existing assets, capabilities under construction, and unproven hypotheses.
Web runtime
OperatingCloudflare Worker, D1, auth, creator, moderation, and payment contracts exist in source.
AI runtime
Not enabledThe current runtime has no production Agents SDK, Vectorize, or agent workflow.
Truth layer
In designClaims, source snapshots, verification, approvals, and provenance are not yet migrated.
Git foundation
Open gateA baseline commit exists; remote, clean-clone CI, and inventory remain open.
Investor readiness
P8Data room, financial receipts, and cohort dashboards open only with real evidence.
THESIS
We are not competing on article volume. We are competing on traceable truth.
City-discovery products are typically strong in maps, booking, reviews, or editorial media. Tram Saigon organizes these needs around a different principle: every sensitive fact should have a source, verification time, expiry, and accountable approval.
Trust architecture
Every public claim links to a source snapshot, freshness, confidence, and approval.
Local intelligence
Neighborhood rhythm, oral history, fieldwork, and lived experience become a structured city graph.
Governed agents
AI retrieves, summarizes, and assists creation; people approve publishing, external actions, and transactions.
Reusable engine
Core truth, approval, connector, and evaluation systems may serve other verticals after Saigon reaches product-market fit.
MARKET FRAME
Start with one city, not an imaginary TAM.
Market sizing will be built bottom-up from users, businesses, creators, and signed contracts. No dollar estimate is published before independent research and conversion evidence.
Ho Chi Minh City
People living, arriving, working, creating, or operating a business in the city.
Vietnamese cities and international users
Expansion of proven taxonomy, source networks, and enterprise use cases.
World Cities
City intelligence, governed agents, and infrastructure APIs for urban verticals with similar trust needs.
Required gate: third-party market research, willingness to pay, cohort retention, CAC, gross margin, and signed enterprise pipeline.
CITY ENGINE
AIOS below. City Engine in the middle. Saigon as the first vertical.
Tram Saigon should not build an isolated AI stack. The web layer owns identity and product policy; the City Engine owns the truth graph, agent workflows, and approvals; model access is connected only after contract and tenant-isolation evidence.
Experience
Search, Concierge, Creator Studio, business discovery, and investor dashboard.
Saigon vertical
Local sources, taxonomy, creator graph, business trust graph, and city-specific evaluations.
Tram Saigon City Engine
Truth graph, retrieval, approvals, media provenance, connectors, ledger, and evaluation.
AIOS / governed infrastructure
Identity, stateful agents, workflows, model governance, audit, payments, and rollback after each dependency is proven.
COMPETITIVE MAP
The competitive map is a hypothesis to validate, not a superiority claim.
| Category | Common strength | Gap to validate | Tram Saigon direction |
|---|---|---|---|
| Google Maps | Places, navigation, and near-purchase intent | Deep editorial context and claim-level provenance | City context linked to sources, neighborhood rhythm, and approval |
| Tripadvisor | Global review scale and discovery | Review consistency and freshness | Verified contributors, editorial gates, and a trust graph |
| Klook | Booking and activity distribution | Urban life beyond transactions | Operating guide first; disclosed commerce later |
| Time Out / Lonely Planet | Editorial brand and destination content | Structured data, workflows, and personalization | Local corpus, claim graph, and creator system |
| Atlas Obscura / Culture Trip | Discovery and storytelling | Operational freshness and local utility | Narrative with practical context and expiry |
| InterNations / Nomad List | Community or data for defined user groups | Local depth and cross-audience governance | One city graph for residents, newcomers, creators, and businesses |
| ChatGPT / general AI | Conversational interface and fast synthesis | Canonical sources, local accountability, and action approval | RAG over approved material, mandatory citation, and evidence-based refusal |
| Wikipedia | Open reference knowledge | Practical workflows, freshness, and contextual service | Reference translated into audited decision support |
Before the data room opens: complete a feature audit, pricing audit, traffic benchmark, and customer interviews for each category.
DATA MOAT
The compounding asset is a city that can be queried and verified.
A moat exists only when data has rights, lineage, freshness, and a correction process. Data volume alone is not defensibility.
Oral history corpus
Consented interviews, transcripts, source IDs, and locality review.
Verified city graph
Claims, places, people, events, neighborhoods, and linked evidence.
Neighborhood ontology
Rhythm, cost, access, work patterns, and change over time.
Creator graph
Topics, locale, score history, corrections, and lived-experience context.
Business trust graph
Claim ownership, verification, operating details, corrections, and disclosure.
Evaluation corpus
Real questions, expected citations, freshness tests, and refusal benchmarks.
NETWORK EFFECT
Every loop must improve the data, not only increase traffic.
- ReadersAsk questions and expose real knowledge gaps.
- CreatorsAdd experience, sources, and accountable corrections.
- BusinessesConfirm operating information and respond to real needs.
- Truth graphImproves coverage, freshness, and governed confidence.
- Better agentsProvide more useful answers with citations and appropriate refusal.
- Repeat usageCreates retention, membership, and additional quality signals.
Engagement is never the sole proxy for truth. Ranking separates relevance, verification, sponsorship, and editorial judgment.
BUSINESS MODEL
Six revenue lines, opened in trust-first order.
Membership
Deeper guides, saved city tools, and structured planning.
AI Concierge Pro
Higher quota, consented project memory, and planning workflows; not more accurate answers for sale.
Creator economy
Paid collections after sustained editorial scores; transparent payout and fees.
Business visibility
Verified listings and contextual placement with sponsor disclosure.
Enterprise API
Licensed, freshness-aware city data and decision-support contracts.
City Engine
Replication or white-label only after the Saigon vertical reaches product-market fit.
UNIT ECONOMICS
The financial model starts with measurable variables, not valuation.
Consumer
ARPU, activation, paid conversion, 90/180/365-day retention, and support cost.
AI
Inference cost per answer, cost per verified artifact, cache hit, fallback rate, and gross margin.
Creator
Payout ratio, review cost, revision count, paid conversion, and creator retention.
Business
Annual contract value, verification cost, lead usefulness, renewal, and churn.
Enterprise
ACV, onboarding cost, data licensing cost, SLA burden, and sales cycle.
Growth
CAC by channel, referral share, payback period, and LTV/CAC after cohorts mature.
Do not publish LTV from immature cohorts. AI, creator payouts, and verification costs remain inside gross-margin calculations.
EXECUTION
From source foundation to Investor Readiness.
- P-1Git, source of truth, remote, and clean-clone CIOpen
- P0Characterization contracts and safe monolith separationNot started
- P1Truth schema, content remediation, canonical ingestionNot started
- P2Semantic search with citation, freshness, and revocationNot started
- P3Read-only Concierge, kill switch, and cost capNot started
- P4Creator Studio, localization, and Editor GateNot started
- P5Image and video workflows with rights and provenanceNot started
- P6Business actions, connectors, and approved procurementNot started
- P7Enterprise API, City Engine, and replication pilotNot started
- P8Due diligence room, IP/data registry, cohort and financial receiptsInitialized by this page
A phase closes only when source, CI, deployment, live behavior, and receipts agree.
CAPITAL PLANNING TOOL
Model capital allocation, not financing terms.
Change the hypothetical raise size to inspect an allocation scenario. Percentages are planning assumptions and must be replaced by an approved operating plan.
Product, platform, and data
Fieldwork, editorial, and source acquisition
Go-to-market and partnerships
Trust, security, and compliance
Operations and contingency
Results are illustrative arithmetic only. They do not represent a valuation, securities offering, use-of-funds commitment, or Founder-approved budget.
P8 · DUE DILIGENCE
The data room opens with receipts.
Corporate and IP
Entity documents, cap table, trademarks, book/content rights, and contributor agreements.
Dataset registry
Sources, consent, licenses, lineage, retention, deletion, and geographic coverage.
AI evaluation
Dataset version, hallucination, citation, freshness, refusal, bias, and cost reports.
Security
Threat model, access review, incident process, secret rotation, backup, and restore evidence.
Financial
Bank/provider reconciliation, revenue recognition, unit economics, and monthly close.
Customer cohorts
Activation, retention, conversion, churn, and qualitative research.
Growth
Channel attribution, CAC, referrals, content efficiency, and business lead quality.
Release
Commit, CI, deployment identity, rollback, live probes, and Founder acceptance.
The private data room and investor-relations channel are not yet published. We do not collect investor information through a form without an owner, privacy notice, and retention policy.
Review the public technical roadmap