AI workflows for industrial buyers

Turn a sourcing question into a careful import check.

An AI assistant can map a buyer's real decision to Tandom's live HTS, duty, AD/CVD, PGA, and regulatory-data calls. The useful result is not a confident shortcut. It is a traceable workflow that asks for missing facts, uses current responses, and keeps the limits visible.

Explore the live API →Connect an AI assistant →

A safe agent pattern

1. Define the buyer's decision and known facts.
2. Ask for the facts required by the live call.
3. Run only current REST endpoints or tools returned by MCP discovery.
4. Return the response, assumptions, citations, and decision boundary.

Start with the decision

Name the product, decision, known facts, and assumptions. The agent should ask for missing facts before it calls a rate or screening endpoint.

Compare the US import-cost impact of buying this same component from Mexico and China.

Give the agent the live docs

Use Tandom's LLM-readable reference so the agent works from current inputs, response fields, and limits.

https://docs.tandom.ai/llms-full.txt

Use REST or discover MCP tools

A coded integration can call REST. An MCP client should call tools/list and use only the tools returned for that connection.

POST https://mcp.tandom.ai/mcp { method: "tools/list" }

Copy into an AI assistant

Make missing facts part of the workflow.

This prompt tells the assistant to use the current docs, stop when facts are missing, and report actual response fields instead of manufacturing a polished answer.

You are helping a US industrial buyer make a sourcing or import decision.

Read Tandom's current API reference:
https://docs.tandom.ai/llms-full.txt

Buyer decision:
<describe the real product, suppliers, countries, and decision>

Rules:
1. Use only live endpoints in the current docs or MCP tools returned by tools/list.
2. Separate confirmed facts, assumptions, and missing facts.
3. Do not invent HTS codes, product facts, duty rates, AD/CVD matches, PGA requirements, citations, or supplier claims.
4. Ask for required facts before calling an endpoint. If the facts remain insufficient, stop and explain why.
5. Quote material values and citations from the actual response. Distinguish a search candidate or advisory signal from a determination.
6. State what the live call does not cover and when human review is required.

Return:
- the missing facts to collect first;
- the live calls to make, in order;
- the input mapping for each call;
- how to compare the responses without adding unsupported claims;
- the boundary between free self-serve tools and a paid sourcing project.

Buyer workflow library

Five decisions the live surface can support.

These are workflow patterns, not sample results. Rates, cases, agency flags, citations, and freshness must come from the response returned when the buyer runs the call.

Industrial procurement team

Compare origin scenarios before choosing a supplier.

Buyer asks

We have quotes for the same machined component from Mexico and China. We know the 10-digit HTS code, customs value, shipment mode, and expected entry date. Compare the US duty and fee impact by origin. Keep supplier price and freight separate, and tell us what is still missing from a full landed-cost comparison.

Careful agent workflow

1Confirm that the product specification, HTS code, customs value basis, mode, and date are comparable across both quotes.
2Check the regulatory-data status before relying on current tariff data.
3Call duty calculation once per country of origin with the same value, mode, and date.
4Compare only fields returned by Tandom, including tariff layers, MPF or HMF, effective rate, citations, and warnings.
5Keep supplier price, international freight, insurance, inland transport, tooling, and financing as separate buyer-provided inputs.

Live public calls

GET/v1/regulatory-data/statusLive
MCP tool: tandom_regulatory_status. Confirm the freshness and health of the regulatory sources.
GET/v1/duty/calculateLive
MCP tool: tandom_duty_calculate. Calculate the duty and customs-fee component for each origin.

What this can support

A side-by-side comparison of the customs amounts and tariff layers returned for each origin, with the input assumptions kept visible.

Boundary

This is not a complete landed-cost quote. The live call does not supply vendor pricing, freight quotes, tooling, quality cost, payment terms, or supplier capability.

Strategic sourcing team

Screen named candidate manufacturers for AD/CVD risk.

Buyer asks

We are considering three named factories for an industrial component. We have each legal manufacturer name, country of origin, product description, possible HTS code, and expected entry date. Screen each candidate for US AD/CVD exposure before we finalize the RFQ shortlist.

Careful agent workflow

1Keep manufacturer, exporter, country of origin, and country of shipment as separate facts.
2Call the AD/CVD check separately for each named manufacturer using the same detailed product description.
3Preserve the returned scope analysis, match type, rate context, Federal Register citations, and ambiguity warnings.
4Do not turn an HTS-level signal into an in-scope determination or call a candidate clear when product scope remains ambiguous.
5Send unresolved scope or identity questions to qualified human review before purchase or entry decisions.

Live public calls

GET/v1/adcvd/checkLive
MCP tool: tandom_adcvd_check. Screen a described product and named manufacturer or exporter against current AD/CVD orders.

What this can support

A candidate-by-candidate risk screen that carries forward Tandom's returned matches, citations, rates, assumptions, and unresolved scope questions.

Boundary

AD/CVD screening is not manufacturer verification, sanctions screening, financial diligence, quality validation, or a legal scope ruling.

Buyer and engineering team

Find the missing product facts before HTS and duty work.

Buyer asks

A supplier described the item as a stainless pump assembly and suggested HS 8413. Before anyone chooses a 10-digit US code or estimates duty, give us the product questions we need the supplier and engineering team to answer.

Careful agent workflow

1Separate confirmed facts from supplier shorthand, assumptions, and missing facts.
2Ask only classification-relevant questions, such as principal function, materials, construction, power source, flow or pressure characteristics, components, and whether the item is a complete machine or part.
3Collect country of origin, customs value, expected entry date, mode, and any special material content before requesting a duty amount.
4After the facts arrive, search HTS candidates and read the relevant hierarchy and legal notes. Treat search results as candidates, not a final classification.
5Call duty calculation only after a human-reviewed code, origin, value, and date are available.

Live public calls

GET/v1/hts/searchLive
MCP tool: tandom_hts_search. Find candidate codes from the improved product description.
GET/v1/hts/hierarchyLive
MCP tool: tandom_hts_hierarchy. Read the hierarchy around a candidate code.
GET/v1/hts/notesLive
MCP tool: tandom_hts_notes. Fetch the legal notes that may control the classification.
GET/v1/duty/calculateLive
MCP tool: tandom_duty_calculate. Calculate duties after the required facts and reviewed code exist.

What this can support

A fact-request checklist followed, only when enough information exists, by documented HTS candidates and a duty result for the reviewed code.

Boundary

The live HTS endpoints search and explain the tariff schedule. They do not replace a defensible classification analysis or invent missing technical facts.

Import and supplier-quality team

Check PGA and document readiness before the first order.

Buyer asks

We selected a supplier and have a reviewed HTS code, product description, packaging details, and a list of documents the factory can provide. Before the first order, identify likely US Partner Government Agency flags and compare the returned document requirements with our list.

Careful agent workflow

1Confirm that the reviewed HTS code and product description match the actual first-order configuration.
2Call the PGA check with the product description and include the wood-packaging fact when relevant.
3Compare the returned agency flags and required-document fields with the buyer's supplier-document list.
4Label missing, expired, product-mismatched, and still-unverified documents separately.
5Escalate admissibility, agency registration, testing, labeling, or filing questions to the responsible specialist before the order ships.

Live public calls

GET/v1/pga/checkLive
MCP tool: tandom_pga_check. Identify likely PGA and ACE tariff flags plus the associated document signals.
GET/v1/regulatory-data/statusLive
MCP tool: tandom_regulatory_status. Check source freshness before relying on the result.

What this can support

A first-order review list showing returned agency flags, document signals, buyer-supplied evidence, and unresolved items.

Boundary

The endpoint identifies tariff-based PGA signals. It does not validate document authenticity, certify compliance, or make a final admissibility decision.

Buyer with an active requirement

Know where free tools stop and managed sourcing begins.

Buyer asks

We have a drawing and specification but no approved supplier. Can the Tandom API find and qualify manufacturers, run the RFQ, manage samples, and get us ready for a first order?

Careful agent workflow

1State the boundary before proposing calls: the public API supports import-intelligence research when the required product, HTS, origin, value, and supplier facts exist.
2Use the live HTS, duty, AD/CVD, and PGA calls only for the decisions they actually support.
3Do not claim that a public endpoint searches for factories, verifies capability, runs an RFQ, manages samples, or approves a supplier.
4Route an active supplier-development requirement to Tandom's paid sourcing service for managed execution.
5Keep final technical, quality, commercial, and purchasing approval with the buyer.

Live public calls

GET/v1/hts/searchLive
MCP tool: tandom_hts_search. Support self-serve tariff research once product facts exist.
GET/v1/duty/calculateLive
MCP tool: tandom_duty_calculate. Quantify duties and customs fees for a defined scenario.
GET/v1/adcvd/checkLive
MCP tool: tandom_adcvd_check. Screen a described product and named candidate for AD/CVD risk.
GET/v1/pga/checkLive
MCP tool: tandom_pga_check. Identify likely PGA and document signals for a reviewed code.

What this can support

A clear split between self-serve import intelligence and the managed supplier-development work the requirement still needs.

Boundary

Manufacturer search, capability screening, RFQ management, quote normalization, samples, follow-up, and first-order planning belong to a paid sourcing project, not a free API call.

Discuss a sourcing project →

Product boundary

Free import intelligence. Paid sourcing execution.

Use the free tools and API

Search the HTS, inspect hierarchy and notes, calculate duties and customs fees, screen a described product for AD/CVD exposure, check PGA signals, and confirm regulatory-data freshness. The buyer or advisor reviews the result and owns the decision.

Review live API functions →

Use a paid sourcing project

Bring Tandom an active requirement when your team needs managed manufacturer search and screening, RFQs, quote normalization, samples, technical and commercial follow-up, or first-order and import planning. Your team keeps the supplier relationship and final approval.

Discuss a sourcing project →

MCP execution

Discover the callable tools for this connection.

REST is a direct fit for a coded integration. MCP lets an AI assistant discover the same read-only functions during a conversation. Start with tools/list, then call only a tool returned by the server. Do not copy a roadmap name into a production workflow.

Open llms.txt →Browse API reference →Install the connector →

POST https://mcp.tandom.ai/mcp
Authorization: Bearer <Tandom API key>
Content-Type: application/json

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/list",
  "params": {}
}

// Choose the next tool from this response.
// tools/call inherits the underlying REST function's inputs and limits.

Choose the path that matches the work.

Use the API for self-serve import intelligence. Connect an AI assistant for conversational tool use. Bring an active supplier requirement to Tandom when the work needs managed execution.

Explore the API →View connectors →Discuss sourcing →

Turn a sourcing question into a careful import check.

You are helping a US industrial buyer make a sourcing or import decision. Read Tandom's current API reference: https://docs.tandom.ai/llms-full.txt Buyer decision: <describe the real product, suppliers, countries, and decision> Rules: 1. Use only live endpoints in the current docs or MCP tools returned by tools/list. 2. Separate confirmed facts, assumptions, and missing facts. 3. Do not invent HTS codes, product facts, duty rates, AD/CVD matches, PGA requirements, citations, or supplier claims. 4. Ask for required facts before calling an endpoint. If the facts remain insufficient, stop and explain why. 5. Quote material values and citations from the actual response. Distinguish a search candidate or advisory signal from a determination. 6. State what the live call does not cover and when human review is required. Return: - the missing facts to collect first; - the live calls to make, in order; - the input mapping for each call; - how to compare the responses without adding unsupported claims; - the boundary between free self-serve tools and a paid sourcing project.

Buyer and engineering team

Find the missing product facts before HTS and duty work.

Buyer asks

Careful agent workflow

1Separate confirmed facts from supplier shorthand, assumptions, and missing facts.
2Ask only classification-relevant questions, such as principal function, materials, construction, power source, flow or pressure characteristics, components, and whether the item is a complete machine or part.
3Collect country of origin, customs value, expected entry date, mode, and any special material content before requesting a duty amount.
4After the facts arrive, search HTS candidates and read the relevant hierarchy and legal notes. Treat search results as candidates, not a final classification.
5Call duty calculation only after a human-reviewed code, origin, value, and date are available.

Live public calls

GET/v1/hts/searchLive
MCP tool: tandom_hts_search. Find candidate codes from the improved product description.
GET/v1/hts/hierarchyLive
MCP tool: tandom_hts_hierarchy. Read the hierarchy around a candidate code.
GET/v1/hts/notesLive
MCP tool: tandom_hts_notes. Fetch the legal notes that may control the classification.
GET/v1/duty/calculateLive
MCP tool: tandom_duty_calculate. Calculate duties after the required facts and reviewed code exist.

What this can support

A fact-request checklist followed, only when enough information exists, by documented HTS candidates and a duty result for the reviewed code.

Boundary

The live HTS endpoints search and explain the tariff schedule. They do not replace a defensible classification analysis or invent missing technical facts.

Discover the callable tools for this connection.

POST https://mcp.tandom.ai/mcp Authorization: Bearer <Tandom API key> Content-Type: application/json { "jsonrpc": "2.0", "id": 1, "method": "tools/list", "params": {} } // Choose the next tool from this response. // tools/call inherits the underlying REST function's inputs and limits.