Products

In-Equal builds the factory data layer for Chinese manufacturing: we reconstruct what 4.8 million factories actually do from fragmented public data, then deliver it three ways — a platform you can use directly, an API you can call, and AI capabilities that get wired into a company’s own systems.

Tianxia Gongchang · the platform

Search factories directly on web, mini-program, and app. Search, factory detail, decision-maker contacts, and conversational AI search all run on the same data and the same engine, with three uses through one entry point:

• Find customers — filter by industry, geography, capacity, export activity, and contact quality to get today’s call list
• Find suppliers — find plants that can hold a spec, compare across an industrial-belt cluster, and check credentials, scale, and operating status before reaching out
• Look up a company — type a company name and see its factory profile and contacts

Data scale:

• 4.8 million factory profiles with depth dimensions: products, capacity, customer structure, equipment level — beyond registry basics
• 13.95 million anonymized decision-maker phone numbers + 5.6 million WeChat IDs
• 1.08 million factories with verified export business
• 1,000+ industrial belts, including cross-city and cross-province township clusters

Learn more about Tianxia Gongchang →

Open platform · MCP and API

The same capabilities, exposed to developers and AI clients. MCP and REST share one auth, metering, rate-limit, and audit path.

Five capabilities are open today: factory search, factory detail, company contacts, factory deep-dive research, and natural-language factory search. Search intent takes three values — sales (find customers), purchase (find suppliers), and company_query (look up one company) — the sell side and the buy side hit the same index and the same fields.

Metered per call, with per-call usage and billing visible in the console. Docs and sandbox keys at the Tianxia Gongchang open platform.

Custom enterprise AI

27 capabilities across seven job families. They read two layers — the national factory data layer and the company’s own orders, inventory, and documents — and write results back into existing systems.

A few of them:

• AI procurement — send one requirement to dozens of plants, collect quotes, and return a single recommendation; score suppliers on on-time delivery and pass rate, and flag registry or litigation changes the day they appear
• AI sales — dial a list, qualify intent on the call, and hand back a card of the key points; apply a cost model plus the day’s material prices to produce a quote ready to send
• AI export — screen customs records and trade-show directories for real buyers and fill in decision-maker emails and phones; check export controls, sanctions lists, rules of origin, and environmental/labor requirements
• AI market intelligence — track newly registered, newly built, and newly commissioned plants nationwide and push a same-day list; sweep tender notices and filter them by category

This line is delivered as custom work rather than off-the-shelf software: a business diagnostic first, one or two high-value scenarios piloted, then custom development, launch, and monthly review. Full capability list at the AI Agent marketplace.

Why this is hard

Factory data is unlike consumer data — no public APIs, no aggregated ratings, no review sites. Building this layer requires solving five things at once:

1. Data acquisition — factory websites, patents, hiring posts, bid records, trade-show directories, corporate PR; every source has to be scraped in-house
2. Data cleaning — registered legal entities and actual production entities are often separate (listed parent vs tier-three subsidiary, parent vs site company, brand owner vs OEM)
3. Sub-industry classification — standard industry codes are too coarse; long-tail township clusters need non-standard classification (e.g. Dainan stainless steel, Dongtai stainless steel, Jingjiang shipbuilding, Hangji toothbrushes — none mapped cleanly in standard four-digit codes)
4. Cross-region clustering — township clusters routinely span cities and provinces (Dainan-Shiyan spans Taizhou-Yancheng, Shengze-Puyuan spans Jiangsu-Zhejiang) — administrative boundaries don’t work
5. AI-agent reconstruction — turning the raw data above into answerable questions: what is this plant producing today, how much can it make, who signs off, is it worth contacting

In practice the most valuable asset is the data layer itself, plus the ways of calling it that grew around it.

What’s next

The next steps center on links in this chain that still aren’t tooled — factory-activity alerting, cross-tool data flow, registry-based verification of credentials and certifications. Specifics will be announced when they ship.