Recipe

Migrate from scraping to the API

If you scrape Credicorp figures today, map each value to an endpoint and cut over. Products, pricing, quotes and loyalty are all published — the API is accurate, stable and won't break on a layout change.

2 min read

MapValue → endpoint

Cut overThen delete scraper

StableNo layout breakage

Map scraped values to endpoints

Everything a scraper pulls from the marketing site is published on the public ring: product names and descriptions from products, figures from config/pricing, live quotes from quote/onetime, and the ladder from loyalty/tiers. Build a mapping table from each scraped field to its endpoint.

Cut over and delete the scraper

Because these endpoints are the single source the site itself reads, the API values reconcile exactly with what you were scraping — so cut-over is low risk. Replace each scrape with its API read, verify parity, then delete the scraper. You lose the brittleness (a layout change no longer breaks you) and gain freshness.

For agents, prefer MCP

If your consumer is an AI agent, skip HTTP mapping and register the MCP server — the tools hand the model authoritative figures directly. See building an assistant. Credicorp's ai.txt welcomes this and asks only for attribution.

Frequently asked questions

Will API figures match what I was scraping?

Yes. The public endpoints are the single source the marketing site reads, so the values reconcile exactly. That makes cut-over low risk — verify parity, then retire the scraper.

Is scraping even necessary any more?

No — everything a scraper pulls (products, pricing, quotes, loyalty) is published on the public API, which is accurate and immune to layout changes. The API is strictly better than scraping.