The AI Epoch
THE AI EPOCH | EPISODE 2: BEYOND THE CHATBOT—BUILDING RELIABLE AI AGENTS WITH MANJIT JOHAL In this episode, host Joe Watkins welcomes Manjit, a 3-time founder and veteran architect (formerly of Ocado and HBO), to discuss the shift from experimental AI pilots to enterprise-grade automation. As a co-founder of Kritmatta, Manjit shares how his team is solving the "trust gap" by building AI agents that are deterministic, auditable, and grounded in real-world business context. KEY TAKEAWAYS FROM THE CONVERSATION: * The Problem with "Prompt Chains": Why traditional chatbots often fail in business settings due to hallucinations, and how Kritmatta uses fixed execution graphs to ensure reliability. * The System of Context: A look at the technical challenge of "entity resolution"—connecting a stray WhatsApp message to a specific HubSpot deal and an email thread to create a unified knowledge graph. * SME-First Automation: Why small to medium enterprises (starting with recruitment firms) are the "wedge" for AI adoption, as they often lack the engineering teams to build these complex systems themselves. * Multi-Agent Situational Awareness: A peek into the roadmap where multiple agents across different channels (LinkedIn, Calendar, CRM) share intelligence to assemble insights automatically without being asked. * The "Boring" Rule of Success: Manjit’s advice for business owners: don't start with a grand vision. Instead, pick one "boring," repetitive process, put a human in the loop, and scale only once trust is earned.
3 episodios
Comentarios
0Sé la primera persona en comentar
¡Regístrate ahora y únete a la comunidad de The AI Epoch!