How we automated technical support for 1,500+ complex products

Our client, a global leader in development systems and modules for embedded electronics, faced the problem that shadows every fast-growing engineering business: exponential support demand.

With a portfolio of more than 1,500 active products—each with its own datasheet, code libraries, and specific user instructions—the volume of information became a barrier to efficient operations.

Before the Asarum AI implementation, the engineering team was the first line of defense for technical requests. The analysis showed alarming results:

• 30% of engineering time was spent searching internal documentation to answer repeated customer questions.
• Response lead time: due to query complexity, customers waited up to 48 hours for precise technical answers.
• R&D stagnation: instead of building new products, senior engineers acted as “living search engines.”

We did not want a basic chatbot. We built an intelligent engineering assistant that deeply understands technical documentation.

Key system capabilities:

1. Structured chaos: the AI agent was trained on 50,000+ pages of technical documentation, thousands of lines of code, and a decade-long archive of forum discussions.
2. RAG technology (Retrieval-Augmented Generation): the system does not guess. It retrieves the exact data and generates responses with citations, eliminating hallucination risk.
3. 24/7 multilingual support: with a global footprint, the agent supports users in all time zones and languages.

The Asarum implementation transformed operational efficiency:

• 70% faster responses: queries that used to take hours are now resolved in seconds.
• Back to R&D: engineers recovered 12+ hours per week per person, directly accelerating product release cycles.
• Scalability without headcount: the client increased supported products by 20% in the first year without hiring additional support staff.