Synopsis: In this Techstrong.ai Leadership Insights video, Boris Bialek, vice president and global field CTO for Industries at MongoDB, dives into the some of the example of successful artificial intelligence (AI) projects that can be found in the newly published book, "Architectures for the Intelligent AI-Ready Enterprise: Building real-world solutions with MongoDB".
Bialek says the AI conversation was flooded with hype and “marketing fluff,” while real implementation guidance was in short supply. The book, now a hefty 500 pages, distills architectures and patterns from actual customer projects, not slideware. Those projects span industries but share a common thread: using data and architecture, not just models and prompts, to move the business needle.
He highlights a cancer research institute in Paris using AI to improve personalized treatment strategies by learning from hundreds of thousands of cases, and a collaboration with Center Reach to support families with autistic children. On the commercial side, he points to a major apparel brand that upgraded search from simple keyword queries to multimodal, multilingual discovery — like snapping a photo of a celebrity’s sneakers and finding comparable options in 50 languages.
Bialek returns repeatedly to a sobering stat: Roughly 95% of AI projects fail to deliver expected value. In his view, many organizations start with the LLM and prompt engineers and only later realize they lack foundations for data retrieval, embeddings, latency, and context. He argues the real work is in “context engineering,” agentic memory, and making data fit-for-purpose in real time — especially as multi-agent systems proliferate.
The book also tackles governance and cost: tokenizing sensitive data, tracing lineage for auditors, orchestrating multi-agent workflows, and caching rote responses to avoid runaway token bills. Structured by industry but designed to be cross-applicable, the book incorporates architectures from partners and clients, not just MongoDB, to give enterprises reusable patterns they can adapt to their own AI journeys.
