The Missing Thread In Product Development
Venkatesh Sankaran is the Founder & CEO of Saguna Consulting Services, turning AI, cloud & digital transformation into business outcomes.
gettySooner or later, many product teams hear the same question from the market: Why didn’t this work? They can explain what happened, but often cannot show which belief, design choice or trade-off led to it. AI can help a team build the wrong product faster, and with even less record of why.
After three decades in consulting, product development, technology transformation, go-to-market and more recently, AI-led innovation, I have seen one pattern persist: Learning gets lost in the handoffs. Nowhere is this more evident than across the product development lifecycle (PDLC), from initial concept to commercial outcomes.
As AI becomes embedded in every stage, from research synthesis to production monitoring, the competitive advantage is shifting from how quickly teams build to how effectively they learn. What matters is not just accelerating execution, but preserving the reasoning behind decisions, connecting assumptions to outcomes and applying those insights to build better products.
A product moves through five stages: idea mapping, experience architecture, product development, deployment and commercialization. Yet the decisions shaping its success begin even earlier, with the investment itself. Which problem deserves funding, what returns are expected and how success will be measured are often agreed upon in planning meetings, only to lose visibility as the product moves forward.
This disconnect continues across the lifecycle. Critical assumptions, design choices and market insights rarely travel with the product from one stage to the next. Much of that context remains with individuals rather than becoming part of a shared record, making it vulnerable to team changes and shifting priorities.
AI has made this gap more consequential. Many teams begin where the gains are most visible: writing code. But the decisions that determine what to build, for whom and why receive far less attention. The result is faster development without necessarily improving the quality of the decisions driving it.
The consequences of this disconnect extend beyond product development to business performance. McKinsey’s The State of AI in 2026 report points to a related gap: About 80% of respondents said AI improved their personal productivity, yet only 37% reported any EBIT impact. Roughly 6% were high performers, attributing at least 5% of EBIT to AI, and they were far more likely to have redesigned workflows around it. Productivity is easy to feel; business value is harder to establish.
Our experience with a healthcare company brought this challenge into sharp focus. We worked on a patient care application that used an AI assistant to support patients between consultations. Technically, the solution performed well. The assistant passed its tests, clinicians validated its responses and the development teams met their targets. Yet four weeks after onboarding, only about 25% of new patients remained active. The investment in AI was delivering functionality, but its contribution to patient engagement remained unclear.
The problem extended beyond the technology. The initiative had started with a broad objective of improving patient engagement, but without a measurable outcome connecting product decisions to patient behavior. Each team had optimized its own responsibilities, while the assumptions behind those decisions remained scattered across functions.
We shifted the focus to measurable product outcomes, establishing a target of approximately 35% four-week retention. Using AI to analyze patient conversations alongside application usage patterns, we identified a critical disconnect: The assistant was drawing on information captured during registration, while clinicians had updated patients’ care plans. The technology was functioning as designed, but the information it relied on no longer reflected the patient’s latest clinical context.
The solution was not another AI feature. We introduced a shared context layer that connected product objectives, decision rationale and current care information across clinical, product and engineering teams. Four-week retention increased to approximately 36%, an improvement of 11 percentage points, or 44% relative to the baseline.
The experience reinforced a lesson we have seen repeatedly: AI can improve how individual components perform, but sustainable product value depends on how effectively decisions, information and outcomes remain connected across the lifecycle.
What we uncovered in healthcare was more than a disconnect between an AI assistant and a patient’s care plan. It exposed a larger vulnerability: A product can meet every technical milestone and still drift away from the business objective that justified its investment.
That begins with keeping the reasoning behind the investment connected to every decision that follows. Not through a monolithic AI platform, but through interconnected layers that preserve context as the product evolves.
At the idea-mapping stage, AI can challenge the investment assumptions against market intelligence and customer needs. As the concept moves into experience architecture, those insights inform design choices, which can then be tested against actual user behavior. During development, the same context follows the product through requirements, code and testing.
Once deployed, AI can connect usage patterns, performance issues and customer feedback to earlier decisions. At commercialization, adoption, revenue and customer outcomes can be assessed against the expectations set at the investment stage, helping leaders decide whether to scale, refine or redirect.
The real opportunity is not to make each stage independently smarter, but to ensure that the intelligence gained at one stage informs the next. That is how AI moves from accelerating product development to strengthening the decisions that determine its success.
What if the greatest value of a product launch wasn’t just what it delivered, but what it taught the organization?
With connected context, AI can reason in two directions. Forward, it traces how investment assumptions shape product decisions and where risks emerge. Backward, it connects market outcomes to the decisions and beliefs behind them, surfacing explanations for teams to validate against evidence.
This changes the economics of product development: The return on an investment is not only what the product delivers, but also what the organization learns.
The real risk isn’t building a product that fails. It’s building one, learning nothing and investing in the next with the same assumptions.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
