The Reality of AI in Product Development: What No One Tells You

The Core Argument

Guided by host Sanjay Saini, Product Management and AI Consultant Joel discussed the largely unspoken realities unfolding in enterprise AI. While the speed of AI adoption has shattered historical records, businesses are encountering severe bottlenecks regarding token budgeting, uncontrolled hallucinations, and proper governance. Joel argued that the market narrative around mass "job destruction" is fundamentally flawed. In reality, the industry is experiencing "job repurposing"—eliminating purely entry-level syntax generation while exponentially increasing the demand for human professionals who possess the empathy, negotiation skills, and strategic insight that an LLM fundamentally lacks.

“If you want to drive your company into the ground, especially a startup, give your team unlimited AI usage. The tokens will drive the company into the ground.” Joel (51:00)

Detailed Answers

1. Could you give a brief intro about yourself and your background?

Joel boasts over 14 years of experience traversing project management and business analysis across high-compliance sectors such as banking, airlines, and finance. His recent focus has heavily transitioned to AI consulting and enablement, helping enterprise product managers evangelize and integrate generative AI capabilities specifically within the fast-moving fintech space.

2. How does the adoption rate of ChatGPT compare to platforms like Netflix?

Joel highlighted the unprecedented speed of the current AI transition. It took Netflix roughly three and a half years to reach one million users, and platforms like Instagram took two and a half months. ChatGPT hit one million users in exactly five days. This aggressive adoption speed forces companies to integrate AI urgently, frequently bypassing traditional, long-term corporate governance cycles.

3. Will AI lead to job destruction or job repurposing?

Joel challenged the doomsday narrative of mass unemployment, citing recent data indicating that while 85 million jobs will be affected, 97 million new roles will emerge. He categorized this as "job repurposing." AI handles entry-level syntax production and boilerplate documentation perfectly. Consequently, companies need fewer entry-level coders but exponentially more mid-to-senior level engineers to orchestrate, debug, and securely integrate those AI-generated outputs into complex architectures.

4. Why is human input susceptible to unpredictability in AI tools?

Because AI heavily depends on the training data it's fed, trying to model an AI agent off a specific human (like the "Elon Musk" Grok clone) inherits that human's unpredictability, impulse, and bias. Joel emphasized that humans are creatures of habit but also creatures of impulse—a nuance that makes rigid AI cloning both difficult and ethically precarious.

5. What are the new AI governance and risk management challenges?

Advanced LLMs are double-edged swords. Joel noted that tools currently exist explicitly for the purpose of analyzing an organization's network to hunt for vulnerabilities. If deployed by a bad actor, an enterprise can be easily compromised. Internally, strict corporate guardrails are actively being installed to block employees from feeding sensitive, proprietary client data directly into public chatbot models to mitigate hallucination-driven compliance breaches.

6. How do you manage and restrict token usage inside a corporate environment?

Tokens are the new hidden corporate currency. Joel observed startups draining massive budgets by giving employees unlimited access to frontier models. Successful organizations manage this through strict usage-based subscription caps with middle-layer AI vendors, ensuring employees are formally budgeted a set number of processing tokens (e.g., 100 tokens a month per employee) to stop uncontrolled financial bleed.

7. How is the industry using job repurposing to lay off people?

Sanjay and Joel discussed the uncomfortable reality that some companies are using the "repurposing" narrative to quietly lay off older, senior-level managers who have refused to become AI-aware. It isn't just the bottom of the ladder at risk; senior leadership unwilling to adapt to the new orchestration paradigms are being displaced by professionals who can actively evangelize AI strategies.

8. What are your top tips for business analysts and product owners starting with AI?

Stop observing and start getting your hands dirty. Joel's immediate advice is to begin with basic prompt formulation to speed up routine tasks. For those who aren't technical, use AI agents specifically designed to build *other* agents in plain English. Practice by creating a rudimentary bot to schedule tasks or summarize notes, treating AI interaction as a mandatory new language skill for product management.

Don't get caught on the wrong side of the job repurposing curve. Learn how to securely govern enterprise AI pipelines, manage token budgets, and successfully lead modern delivery teams in our Artificial Intelligence training program.