AI SDLC: Agent-Driven Workflow, Humans as Needed | Viral Bajaria, 6sense Chief Innovation Officer

Viral Bajaria, co-founder and Chief Innovation and Ecosystem Officer of 6sense, joins us to explain the AI SDLC, the operating model he is building at 6sense. With AI SDLC, agents orchestrate the full development lifecycle, running continuously in the cloud and deciding when to pull in humans for architecture reviews, security approvals, and design decisions. The team can now create a fully functional prototype ready to show a customer, integrated into 6sense platform services and using the customer's own data, all built from the initial vision in just 2.5 weeks. Topics covered: * The AI SDLC: agents run requirements, architecture, coding, testing, and deployment end-to-end * The output: fully-functional prototypes ready to show to customers, with auth, RBAC, observability, and their real data, all built from an idea in 2.5 weeks * Agent-initiated workflow: the agent manages the process and decides when to pull in humans * Giving product managers direct access to the codebase: ask the code, not the engineers * The living document: a single context layer pulling from Slack, Jira, Confluence, and product systems * Approvals in the chain: everything recorded — who decided what, when, and why * Observability for long-running agentic workflows: sessions, turns, cost per task * Token cost optimization: intake harness, task classification, model routing. Selecting the right model can make tasks 15x cheaper! * Team structure: the 2-FTE model of product development, and why architecture-level judgment still matters most * The abstraction frame: AI as a process-level shift, not just a tool-level one Chapters: 00:52 The evolution of AI engineering at 6sense 03:49 The AI SDLC: agents orchestrate the full development lifecycle 08:19 From vision to customer prototype in 2.5 weeks 10:55 AI-native engineering defined: agents initiate, humans decide 14:18 AI observability and tokenomics: cost per task and model selection 21:27 2-FTE team structure and why architecture judgment still matters 26:04 Adding abstraction at the process level above the tools — and what comes next Inside the CTO Circle is a podcast for engineering leaders scaling AI in production. Hosted by Shruti Bhat and John Troyer from Snowflake. #AIEngineering #CTO #6sense #AgenticAI #SDLC #EngineeringLeadership #Observability