Triqai

What is Triqai?

Triqai is a transaction enrichment API that transforms raw bank transaction strings into structured merchant data. It enriches transactions with merchant names, logos, categories, and location details so downstream apps and data pipelines can use consistent, clean outputs.

Triqai is designed for developer use via a REST API and a Node.js SDK. It uses machine learning combined with real-time web data to parse transaction text into identifiable entities and return enriched results with confidence scores.

Key Features

• Transaction parsing into structured entities: Converts messy transaction text into structured merchant information (e.g., merchant, category, and other related entities) suitable for app UX and analytics.
• Categorization and spend normalization: Assigns categories, detects subscriptions, and helps correct mislabels to keep budgeting and insights more consistent.
• Location enrichment (store-level when available): Resolves merchants to location details such as store numbers and coordinates to support richer experiences and location-aware workflows.
• Intermediary vs. merchant separation: Detects and returns intermediaries (e.g., payment/marketplace entities) separately from the merchant, and supports transactions with multiple intermediaries.
• Developer-first API & SDK support: Integrates through a REST API or the official Node.js SDK (npm install triqai), with features mentioned for production readiness such as automatic retries and auto-pagination, and TypeScript with type definitions.
• Confidence scores in responses: Includes confidence scores (0–100) for each entity (merchant, location, payment processor, and category) so you can programmatically handle uncertain matches.

How to Use Triqai

1. Get an API key and start with the REST API or Node.js SDK. Use the SDK method or the API endpoint to submit raw transaction strings for enrichment.
2. Receive structured enrichment results. Each response returns enriched entities and confidence scores; you can inspect these in your application logic.
3. Apply thresholds for edge cases. Use the confidence scores to decide what to display to users and when to route results for review or fallback behavior.
4. Persist enrichment outputs (optional). Enriched data is yours to store; you can save it in your database or cache it to reduce repeated API calls.

Use Cases

• Personal finance apps: Enrich user transaction text into recognizable merchant names and categories so users can quickly understand spending and trust the information shown in the app.
• Accounting and expense tools: Automatically classify expenses, normalize vendors, and reduce manual reconciliation time by providing structured merchant and category data.
• Banking and open banking workflows: Standardize transaction data across banks and regions by producing consistent merchant-related fields from raw inputs.
• Fraud and risk systems: Add merchant and location context to help detect anomalies and reduce false positives by grounding transactions in merchant/location metadata.
• Spend analytics dashboards: Build more reliable insights across merchants, categories, and regions by enriching raw transactions into structured, reusable data.

FAQ

What does Triqai return for each transaction? Triqai returns structured data derived from the transaction text, including merchants, locations, categories, payment processor detection, and confidence scores for each entity.

How does Triqai decide what a merchant is? Triqai uses machine learning along with real-time web data to parse transaction text into structured entities, then enriches those entities with additional information such as logos, websites, coordinates, and metadata.

Does Triqai work worldwide? Yes. Triqai supports all countries worldwide and handles local character sets and region-specific payment methods. Coverage is stated as strongest in the EU, US, UK, and ANZ.

What happens if Triqai cannot confidently recognize a merchant? You still receive usable data. Triqai provides category and payment processor detection, location when available, and a best-effort merchant guess with a low confidence score; responses are described as rarely empty.

Can I store the enriched results on my own servers? Yes. Enriched data is yours to keep, and you can store it in your database or cache it for downstream systems and to reduce API calls.

Alternatives

• Rule-based transaction parsing (vendor dictionaries): You can map known vendor strings to merchant and category data using maintained rules. This may be simpler to operate but typically requires ongoing updates and may perform worse on unfamiliar or noisy transaction text.
• Generic OCR/NLP pipelines on transaction descriptions: You could build an NLP workflow to extract entities from transaction narratives. This gives more control but requires more engineering effort to achieve reliable merchant, category, and location enrichment.
• Other merchant/location enrichment APIs: Alternative enrichment services provide similar structured merchant and location outputs. The main difference is usually output schema, coverage by region, and how confidence scoring and intermediary handling are represented.
• End-to-end account aggregation + transaction categorization tools: Some tools focus on grouping and categorizing transactions within an aggregation workflow rather than taking arbitrary transaction strings as input. These may fit teams already committed to that ecosystem but can be less flexible when you need to enrich transactions from multiple sources.