AI-Augmented Product Development: A Responsible Methodology for Human-AI Collaboration in Enterprise Product Discovery
1st IEEE International Conference on Responsible Artificial Intelligence · MelbourneAI has cut building software from weeks to days, but deciding what to build still takes 8–12 weeks of research, reviews and meetings.
Four phases where AI proposes and people decide. Each phase ends at an explicit human gate, and the output goes straight into AI-DLC to be built.
Two enterprise deployments. An 8–12-week cycle done in 2 days (about 20×), with people setting aside 9 of 12 AI suggestions at the first gate.
What the deployment showed
Deployment A: a discovery cycle that usually takes 8–12 weeks, done in a 2-day workshop. Calendar time against focused workshop time; the paper notes part of the gain comes from working in one room.
Per activity in Deployment A: research synthesis 3–4 weeks to 90 minutes; PRFAQ and PRD 2–3 weeks to 4 hours; stakeholder materials 1–2 weeks to 2 hours.
Of 12 opportunity themes the AI surfaced, the product leader kept 3. In Deployment B the team rejected 4 of 7 for context the AI lacked.
Consumer services (12 people, 3 business units) and food services and facilities management (15+ people, 7 business units); both 2-day workshops.
Two days, four phases, a human at every gate
Synthesised thousands of customer feedback signals into sentiment maps and 12 opportunity themes.
The product leader narrowed the 12 to 3 priorities in 90 minutes, a step that usually takes 3–4 weeks.
Where the time went, activity by activity
The paper compares calendar time spread across weeks of meetings and reviews with focused workshop time, and notes that part of the gain comes from working together in one room. The AI's share is the drafting: research synthesis, PRFAQ, prototype code and stakeholder materials.
AI-driven development methodologies have compressed software construction from weeks to days, but upstream product decisions (what to build, for whom, and why) remain a bottleneck. This paper introduces the AI-driven Product Development Life-cycle (AI-PDLC), a four-phase methodology that positions AI as a collaborative partner in enterprise product discovery… while preserving human decision authority at every phase transition. We describe the methodology’s relationship to AI-driven software development (AI-DLC), detail the workshop-based execution model including scoping, facilitation, and tooling, and report on an anonymized enterprise deployment. Initial results suggest 15–25× compression of individual product discovery activities and approximately 20× end-to-end cycle reduction, with participants reporting improved stakeholder alignment. We argue that responsible AI in product development requires methodological design that makes human oversight structural rather than optional.
Read the full abstract
- Conference
- IRAI 2026, Melbourne
- Session
- Responsible AI Applications · track: Responsible Decentralized Agentic Systems
- Presented
- Sep 5, 2026
- Paper ID
- IRAI26-000348
G. R. Ramadurai, “AI-Augmented Product Development: A Responsible Methodology for Human-AI Collaboration in Enterprise Product Discovery,” in Proc. 1st IEEE International Conference on Responsible Artificial Intelligence (IRAI 2026), Melbourne, 2026.
Awaiting publication in the IEEE proceedings. It will be listed on my Google Scholar profile once indexed.
Google Scholar profile ↗