Salesforce's Data 360 team converted roughly 100,000 unstructured Informatica documents into a searchable AI knowledge base in 24 days, beating a 30-day post-acquisition deadline. The resulting Informatica Help Agent resolves 80% of customer questions with 5% human escalation.
"The Informatica Help Agent achieved an 80% resolution rate with only 5% human escalation," wrote Irina Malkova, VP of Product and Success Data, with co-author Alexander Smith on Salesforce's engineering blog on February 11, 2026.
When Salesforce completed its acquisition of Informatica on November 18, 2025, the team was challenged to enable the Informatica Help Agent within 30 days. Informatica's documentation was written for humans, not AI: raw HTML included headers, footers, and navigation menus that degrade retrieval, and the knowledge base had complex versioning with many near-duplicate articles differing only slightly across versions.
The team prioritized a production-grade foundation over every complex detail, reusing established Data 360 patterns for ingestion, chunking, indexing, and retrieval instead of rebuilding. Product versioning ran through prompting and configuration rather than system logic changes, keeping effort on retrieval fundamentals. The system shipped in 24 days, ahead of the 30-day deadline; the first developer-documentation ingestion took about three hours, and later updates ran faster as pipelines stabilized.
Data 360 supplies the data foundation — unifying, standardizing, indexing, and activating unstructured knowledge without custom pipelines. For ingestion, Data 360's "sitemaps" feature crawled Informatica's website into conforming knowledge objects, with Python workflows handling more unique content. Preparation normalized HTML and removed noise while preserving meaning, converting documents into consistent chunks for embedding. Retrieval reused Customer Success chunking strategies with filters and metadata tags added during ingestion to narrow results.
The knowledge base spans roughly 100,000 documents from multiple systems with differing structures and metadata; Data 360 pipelines supported hundreds of thousands of documents despite filtering and refresh-timing limits, worked around through preprocessing and configuration. Because pipelines and infrastructure follow a standard design, Informatica-specific tuning stayed localized — no systemic changes or ground-up redesign. The team estimates a rebuild would have required a much larger team and months of extra work; reuse instead delivered equivalent outcomes in a fraction of the usual time while establishing a scalable foundation for future agents.
Salesforce frames the work as a reusable foundation rather than a one-off: the same Data 360 patterns now support all agents augmenting its Customer Success motion, including those on help.salesforce.com and slack.com/help, balancing tailored answers per agent against a durable data layer that can power future agents.