Renso Trade Intelligence Stack

Frank’s trade-data aggregator architecture and intelligence tooling, built as independent side-projects that double as proof-of-concept for the Phase 3 intelligence deliverables in [[renso-engagement-frank-cto]]. Source files live at /Users/franknguyen/renso/; they are working code/specs, not polished deliverables.

See also [[Financial Data Sources]] for the broader context of trade data sources Frank has catalogued.

Architecture Overview: Three-Layer Triangulation

The intelligence stack is organized as three layers that combine macro volume truth, micro entity truth, and an entity-resolution joining layer.

Layer 1: Macro (Volume Baseline)

“What is moving, at what volume, in what HS code, between which country pairs?”

Source	Coverage	Granularity	Access
StatCan Table 34-10-0012-01	Canadian imports by origin	HS-10 imports, HS-8 exports	Free REST API, no key required
US Census Bureau International Trade API	US imports/exports	HS-10, by state/port/district	Free with dev key
UN Comtrade Plus	Global bilateral flows	HS-6	500 req/day free tier
Eurostat Comext SDMX	EU bilateral flows	High granularity	100% free; bulk download available

Lag: All macro sources report at ~45-day lag from month-end. (source: /Users/franknguyen/renso/Master_Trade_Data_Directory.md)

Layer 2: Micro (Entity / Manifest)

“Who specifically is shipping what, to whom?”

Peru SUNAT manifests (“Factory Identification Oracle”): Peru is one of the few countries that makes detailed Bill of Lading data freely accessible. Basic CAPTCHA only. Frank’s implementation scrapes by importer/exporter name or HS code, returning date, FOB value, weight, shipper, and product description. (source: /Users/franknguyen/renso/Product_Intelligence_Engine.py)
Port arrival logs: Vancouver and Tacoma port arrival data
ImportYeti: Scraping US Bills of Lading data — supplier/customer tables mapped back to HS codes from Census/StatCan (source: /Users/franknguyen/renso/Master_Trade_Data_Directory.md)

Layer 3: Joining (Entity Resolution)

“Resolve every scraped company name to a canonical legal identity before joining.”

GLEIF Golden Copy — 4M+ LEIs (Legal Entity Identifiers) indexed locally
Matching via RapidFuzz (fuzzy string matching) or Elasticsearch
Every scraped company must resolve to an LEI before joining macro and micro layers

(source: /Users/franknguyen/renso/Global_Trade_Aggregator_V1_Staff_Engineered.md, source: /Users/franknguyen/renso/Master_Trade_Data_Directory.md)

Mirror Data Technique for Canada

Canada suppresses vessel manifest data (unlike the US, which releases it via FOIA/CBP). The workaround: query Vietnam’s export manifests instead.

Any shipment leaving Vietnam for Canada lists the Canadian consignee on the Vietnamese export side. Vietnam-side data is accessible via SUNAT-style queries and bilateral trade datasets.

This technique has been applied to Renso’s own supply chain — mock data in Product_Intelligence_Engine.py names “Renso Group JSC” as a shipper of Vifon noodles and coconut water from Vietnam, suggesting Frank has already externally fingerprinted Renso’s Vietnamese sourcing before signing any agreement. (source: /Users/franknguyen/renso/Product_Intelligence_Engine.py)

Product Intelligence Engine

File: /Users/franknguyen/renso/Product_Intelligence_Engine.py

A Python class (ProductIntelligenceEngine) implementing the full supply-chain reverse-engineering chain:

Flow: brand → legal owner → factory → global logistics

Layers implemented:

Identity Layer (find_brand_owner): USPTO/trademark lookup to find the parent company from a brand name
Manufacturing Layer (fcc_id_lookup): FCC ID database search for electronics manufacturing specs (scrapes fccid.io)
Logistics Layer (scrape_peru_customs): Peru SUNAT POST request for shipment mirror data — returns date, FOB value, weight, shipper, product description
Integration Engine (generate_intelligence_report): Chains all three into a single intelligence report per brand

Test data in the mock: “Renso Group JSC” appears as shipper in the Peru SUNAT mock return, with:

Shipment 1: Instant Noodles (Vifon), 2,500kg, $45K FOB, 2024-05-01
Shipment 2: Coconut Water 330ml, 8,000kg, $120K FOB, 2024-04-15

This indicates Frank tested the intelligence engine against Renso’s own supply chain profile as part of pre-engagement research. (source: /Users/franknguyen/renso/Product_Intelligence_Engine.py)

Meta Ad Library Tooling

Files: /Users/franknguyen/renso/ad_library_search.py + /Users/franknguyen/renso/get_token.py

Scrapes the Meta (Facebook/Instagram) Ad Library API to identify Vietnamese-food advertisers actively targeting Canada.

Search terms used: Vietnamese food, catering Vietnamese, phở, bánh mì, bún bò, cơm tấm, Vietnamese catering, đồ ăn Việt

API params: ad_reached_countries: CA, ad_active_status: ALL

Fields returned: ad_id, ad_creation_time, ad_creative_bodies/titles/descriptions, page_name, page_id, advertiser_payment_methods, publisher_platforms

Use case: Competitive intelligence input for the Phase 3 “brand and content intelligence” deliverable in the Renso engagement — identifies which Vietnamese-food competitors are advertising in Canada, what creative they’re running, and how active they are. (source: /Users/franknguyen/renso/ad_library_search.py)

Connection to Renso Engagement

These tools are not formally listed as deliverables in the Renso Engagement Proposal. However they are:

Technical prototypes for Phase 3 intelligence systems — the supply chain intelligence, product opportunity intelligence, and brand/content intelligence deliverables in Phase 3 map directly to what these tools already do
Likely the basis for the Renso Hub data-subscription product — the vision document describes “diaspora demand-signal data sold back to platform tenants”; the ad library tooling and trade flow triangulation are exactly the data pipelines that would feed this product
Pre-engagement due-diligence tools — Frank used these tools to fingerprint Renso’s own supply chain before signing, giving him an independent view of what Renso Group JSC is actually shipping from Vietnam

(source: /Users/franknguyen/renso/Product_Intelligence_Engine.py, source: /Users/franknguyen/renso/Renso-Engagement-Proposal.md)

Broader Trade Aggregator Context

Frank’s trade intelligence work extends beyond Renso into a standalone trade-data aggregator project:

Global_Free_Trade_Aggregator_Spec.md — specification for a broader free-data trade aggregator
Global_Trade_Aggregator_V1_Staff_Engineered.md — staff-engineered version of the aggregator
Trade_Data_Collection_Spec.md — detailed collection methodology

These files live at /Users/franknguyen/renso/ and represent Frank’s broader research into trade data infrastructure. The Renso engagement is both a client and a testing ground for this infrastructure.

See [[Financial Data Sources]] for the existing wiki page that covers some of the same macro data sources (StatCan, EDGAR, FRED) in a broader financial analysis context.

Key Technical Reference: Source APIs

Source	Endpoint	Notes
StatCan WDS	`https://www150.statcan.gc.ca/t1/wds/rest/`	Table 34-10-0012-01 for Canadian imports from Vietnam; no key required
UN Comtrade Plus	`https://comtradeplus.un.org/api/get/{type}/{freq}/{cl}`	500 req/day; Ocp-Apim-Subscription-Key header
US Census Bureau	`https://api.census.gov/data/timeseries/intltrade/imports/hs`	Free dev key; PORT variable for port-level data
Eurostat Comext	`https://ec.europa.eu/eurostat/api/comext/dissemination/sdmx/2.1/data/`	100% free; bulk download for full history
Peru SUNAT	`http://www.aduanet.gob.pe/cl-ad-itconsaduanas/IAduanaS01Alias`	POST request; basic CAPTCHA; free mirror data
Meta Ad Library	`https://graph.facebook.com/v19.0/ads_archive`	Requires Meta app credentials; `ad_reached_countries: CA`

(source: /Users/franknguyen/renso/Master_Trade_Data_Directory.md, source: /Users/franknguyen/renso/ad_library_search.py)

[[renso-engagement-frank-cto]] — the engagement these tools prototype Phase 3 for
[[renso-foods-canada]] — the company whose supply chain has been fingerprinted
[[renso-strategic-vision]] — the Renso Hub data-subscription product these tools would feed
[[Financial Data Sources]] — existing wiki coverage of macro financial/trade data sources
[[Canadian Government Data Indexing]] — adjacent: StatCan and government API patterns