Naresh Gali | Data, AI, and the Future of Human Potential

#DataSeries | #47 For the last 15 years, if you wanted two pieces of software to talk, you built an API ( Application Programming Interface ). APIs are great, but they are rigid. They are " Dumb Pipes ." You: "Get me the data in Row 5." API: "Here is Row 5." If you want Row 6, or if you want to know why Row 5 is empty, you have to write new code. The API has no brain. We have spent the last decade building "Headless" systems. APIs are the Hands (They execute actions). Databases are the Records (They store facts). RAG is the Memory (It stores knowledge). But who decides ? Until now, you decided. You wrote the code: If X happens, call API Y. That isn't Intelligence. That is just a fancy flowchart. The 2026 Shift: Model Context Protocol ( MCP ) The new standard is MCP . Think of it as the " USB-C " for Artificial Intelligence. Instead of building a rigid pipe for every single tool, MCP provides a standard way for AI Agents to ...

#DataSeries | #46 We used to just watch the data (Dashboards). Now, we act on the data (Agents). The most important shift in 2026 isn't just Generative AI . It is Process Intelligence . The " Context Gap " AI Agents are powerful, but they are blind. They don't know how your business actually runs. As @ Wil van der Aalst has taught us, the truth isn't in the SOPs —it's in the event logs . If you want to fix the business, you have to mine the reality, not the theory. The Solution: The Process Architect We are moving away from passive analysis to active orchestration. The Tool: @ Celonis provides the MRI scan of the business ( Object-Centric Process Mining ). The Action: We use that intelligence to guide AI Agents ( AgentC ) to fix problems, not just report them. The Leadership This requires a new kind of thinking. It's about looking at the flow, not just the function. Inspiring to see leaders like @ Alex Rinke building the platform that connects Proces...

#DataSeries | #45 You wouldn't ship code without Unit Tests . Why are you shipping AI Agents without Evals ? Most teams check if their AI works by "Eyeballing it." Dev: "I asked it 3 questions and it looked good." Reality: It failed on question #4 and leaked PII on question #5. In 2026, LLM-as-a-Judge is the standard. We don't grade the AI manually. We build a Grader Agent to grade the Worker Agent . ⸻ The Framework: RAGAS & DeepEval You need 3 specific metrics to trust your system: Faithfulness: Did the AI hallucinate info not in the source docs? Test: If the Doc says "Price is $10" and AI says "$12", Score = 0. Answer Relevance: Did the AI actually answer the user's question? Test: User asked "How do I return this?", AI answered "Our hours are 9-5." Score = 0. Context Precision: Did the retrieval system find the right document? Test: Did it pull the "Return Policy" or the "Shipping P...