https://www.vishshelar.com Engineering leader writing about Salesforce, Enterprise AI, software architecture, and reliable systems. en-us https://www.vishshelar.com/blog/agent-discovery-and-governance https://www.vishshelar.com/blog/agent-discovery-and-governance Thu, 22 Oct 2026 00:00:00 GMT Discovery is just the first layer. A five-layer framework — discovery, identity, authorization, observability, enforcement — for governing AI agents, and how Microsoft, AWS, Salesforce, and security vendors each cover only part of it. https://www.vishshelar.com/blog/hardest-part-of-enterprise-ai https://www.vishshelar.com/blog/hardest-part-of-enterprise-ai Wed, 21 Oct 2026 00:00:00 GMT Enterprise AI projects rarely fail because the model isn't good enough. They fail because nobody trusted what it was reading. https://www.vishshelar.com/blog/dreamforce-2026-roi-reckoning https://www.vishshelar.com/blog/dreamforce-2026-roi-reckoning Tue, 08 Sep 2026 00:00:00 GMT Salesforce's agentic enterprise pitch is compelling on stage. Whether it's compelling on a purchase order is the story that actually matters this year. https://www.vishshelar.com/blog/how-much-should-you-trust-an-ai-agent https://www.vishshelar.com/blog/how-much-should-you-trust-an-ai-agent Tue, 18 Aug 2026 00:00:00 GMT Agent autonomy shouldn't be all or nothing. It should depend on the risk of being wrong. https://www.vishshelar.com/blog/ai-agents-vs-workflows https://www.vishshelar.com/blog/ai-agents-vs-workflows Mon, 03 Aug 2026 00:00:00 GMT Why sending an email after answering a question doesn't make an LLM an AI agent — and how to actually tell workflows and agents apart.