Episode Summary
In this episode, Bhairav and Alan bring in AI practitioner Sanjay Rakshit to unpack what’s really going on with artificial intelligence – beyond the noise, hype, and fear.
They explore what AI actually is (and isn’t), why no one is truly an “AI expert” yet, and how founders and business leaders should think about AI as a power tool rather than a magic replacement for people.
The conversation ranges from:
- The history and evolution of AI and why it suddenly “caught fire” with ChatGPT.
- How hype distorts reality for founders, investors, and even universities.
- How to differentiate between genuine practitioners and people who are just one chapter ahead of you.
- Why AI augments good practitioners instead of replacing them – and what that means for developers, doctors, and small businesses.
By the end, you’ll be less intimidated by AI, more skeptical of big promises, and clearer on how to use it intelligently in your business.
Why Everyone Is Talking About AI – And Why That’s a Problem
- AI has exploded into public consciousness largely because of tools like ChatGPT that regular people can now use.
- Social media and huge platforms mean anyone can broadcast an opinion, regardless of depth of understanding.
- Historically, new technologies matured largely out of public view – experts experimented, failed, refined, and only then did things go mainstream.
- Now, the experimentation, failures, and half-baked ideas are all happening in public, magnifying hype and confusion.
“Just because someone has the largest platform doesn’t mean they have the largest knowledge.”
A Simple Definition: What AI Really Is
- At its core, AI is about getting machines to perform parts of what the human brain does:
- Perceive (see/hear data)
- Analyze and detect patterns
- Interpret what’s going on
- Decide on an action
- AI has been researched for 50–60+ years:
- Early work goes back to people like Alan Turing.
- Progress was limited by compute power and data availability.
- We went through “AI winters” where funding dried up because the technology couldn’t deliver on expectations.
- The recent surge is driven by:
- Cheap, powerful GPUs
- Large data centers
- Massive, cheap storage
- Abundant data
Opinions vs Facts: Why the AI Space Feels So Noisy
- Research is still catching up; nobody has the full picture.
- Many people online are regurgitating what they’ve read elsewhere.
- Universities too are racing to offer “AI” and “data science” courses, but only a small number are true centers of excellence with deep research foundations.
- Investors (VCs, PE) are not immune: billions have been poured into “GPT startups” that later collapsed.
“What we have today are opinions largely, and very few factual information.”
Tool Users vs AI Companies – A Crucial Distinction
Sanjay draws an important line between two types of organizations:
- Tool Users
- Example: a company that manufactures train wheels and uses AI just to improve its processes.
- AI is like buying power tools at B&Q – it speeds you up, but it’s not your core business.
- If a tool fails or changes, the business impact is limited.
- AI Companies / AI Practitioners
- Example: a company whose competitive edge relies on how it designs and applies AI.
- Here, AI isn’t just a tool; it’s strategic infrastructure.
- Choosing the wrong approach, stack, or architecture can easily waste tens or hundreds of millions.
Founders need to be clear:
- Are you just using AI tools to help your existing business?
- Or is AI part of your core product / differentiation, where you need deeper expertise and architecture thinking?
The AI Stack: Four Levels to Understand
For enterprises, Sanjay outlines four broad layers:
- Infrastructure – GPUs, compute, storage, data centers.
- Model Providers – Those who build and host the base models (e.g., LLMs).
- Tools Providers – Frameworks and orchestration tools (e.g., LangChain, LangGraph, CrewAI).
- Application Builders – The companies and teams turning those into real products.
If you’re building applications on top, bad early decisions at the foundation level can be extremely expensive to unwind.
How to Tell If Someone Really Knows What They’re Doing
There are no perfect filters, but a few practical approaches emerged:
- Ask: “Why did you need AI for this problem?”
This single question often reveals whether AI is genuinely required or just bolted on for buzz. - Interrogate their experience like any other hire or supplier:
- What have you actually built?
- How complex was it?
- Where is it being used?
- Did it succeed? What went wrong?
- Look beyond keywords and hype:
- Recruiters may chase buzzwords.
- Many “AI experts” popped up overnight after ChatGPT, just as “GDPR experts” came out the week the law was announced.
- Apply critical thinking, even if you’re not technical:
- Dig into people’s incentives (e.g., are they fundraising? selling a tool?).
- Don’t treat a viral article or big-name podcast as gospel.
“There’s no harm in being wrong – but you have to have applied critical thinking.”
Lessons from Past Tech Hype: Blockchain, Mobile Apps, GDPR & Windows
The hosts compare AI hype to previous waves:
- Blockchain – Predicted to put everything “on-chain” and kill fiat currencies. Reality: niche but useful, far from the promised revolution.
- Mobile apps – Everyone claimed they can build an app; the way you separated good from bad was portfolio and outcomes, not pitch.
- GDPR – Overnight, the market was full of “GDPR experts” before anyone knew how it would actually be interpreted.
- Operating Systems (NCR example):
- The company did not jump instantly from OS/2 to Windows XP or Windows 10.
- They negotiated extended support to move carefully, because stability for customers mattered more than being on “the latest and greatest.”
The message: you’ve been here before with other technologies. The principles of due diligence and rigor still apply.
AI as a Power Tool, Not a Replacement
A core theme Sanjay returns to repeatedly:
“AI is not a replacement for a practitioner. It’s a power tool to work alongside them.”
- For developers:
- AI can speed up writing boilerplate, suggesting code, and handling repetitive tasks.
- It cannot replace the thinking, architecture, and system design that great engineers bring.
- Early studies claiming “20–30% productivity gains” are now being tempered by reality: developers may spend more time fixing AI-generated code.
- For doctors:
- AI can improve detection rates in some cancer diagnostics.
- But over-reliance can also dull human skills, as some doctors defer too heavily to the AI.
- Across professions:
- AI reduces the amount of low-value work required to reach a given outcome.
- The practitioner’s judgment, domain knowledge, and critical thinking are still absolutely essential.
The Power Tools Analogy: Builders, Houses & Fast Food
Sanjay and Alan lean heavily on a few relatable analogies:
Power Tools & Builders
- Before power tools, it might take two years to build a house.
- With power tools, fewer people can build a house faster, but:
- We didn’t eliminate builders.
- We built more houses, creating more total work and opportunity.
- Similarly, before AI tools, you might need 20 people for a month to get a startup from 0 to 1.
- Today, 2–3 good practitioners with AI tools might get there in a week.
- That’s not just job loss – it also enables more startups, more products, more experimentation.
Fast Food vs Gourmet
- Fast food is not “better” than a slow, gourmet meal – but it absolutely has its place.
- AI allows the equivalent of “fast food” solutions:
- Quick-and-dirty prototypes
- Rapid experiments
- That doesn’t kill the need for deep, careful, high-quality work – it just adds more options.
“Fast food has a market, but it hasn’t killed gourmet.”
The Risk of Feeling Like a Genius Too Quickly
Alan and Sanjay raise a generational and behavioral concern:
- Younger or less-experienced people can feel like geniuses overnight because AI gives them great-looking outputs quickly.
- But if you don’t actually understand the underlying problem, you:
- Won’t know if the answer is any good.
- Can’t reliably adapt or debug what the AI gives you.
- This is especially dangerous for founders building products with tools like Cursor or Lovable:
- It might take 5 minutes to build a first version.
- It can take three days to change it, because you don’t understand the code or architecture.
This reinforces the message: AI can give you a head start, but you still need real expertise if you want to scale and sustain.
Jobs, Startups, and the Bigger Economic Picture
- AI and automation will meaningfully reduce the number of people needed for some tasks.
- But as with power tools and construction:
- You don’t just get fewer builders; you get more building.
- Lower friction creates new businesses, new roles, new opportunities.
- For founders:
- You can do more with less – but only if you invest in good practitioners and smart architecture.
- There will likely be more companies, more experimentation, and more need for people who can ask the right questions and design robust systems.
Why It Feels So Chaotic Right Now
Sanjay’s closing perspective on why AI feels uniquely unsettling:
- The pace and nature of learning in AI is not fundamentally different from prior tech revolutions.
- What is different:
- Everything is now visible.
- We witness in real time: failed startups, blown investments, overhyped launches, and broken promises.
- In the past, many of these failures happened behind closed doors. We only saw the polished products after years of quiet iteration.
- Today, we’re watching the entire messy process live, and that amplifies both fear and FOMO.
“Today we have complete visibility to all the failures. What we see scares us – what we don’t, we don’t care.”
Practical Takeaways for Founders & Business Leaders
- Don’t fall for the loudest voice. Size of platform ≠ depth of knowledge.
- Use AI as a power tool, not a crutch. It should augment capable people, not excuse you from understanding your own business.
- Be clear on your role:
- Are you an AI tool user, or an AI company?
- Your risk profile and expertise needs are completely different.
- Ask better questions:
- Why does this problem need AI at all?
- What outcomes will it improve, and how will we measure them?
- Apply the same rigor you used for past technologies:
- Reference checks, portfolios, pilots, and proof of concepts.
- Don’t migrate everything to the “latest model” just because it’s new.
- Expect a long game.
Just like “digital transformation” has been ongoing for decades, AI will be a continuing journey, not a one-off project.
Guest Bio
Sanjay Rakshit
Sanjay is an AI practitioner with decades of experience in engineering and complex systems. He focuses on helping enterprises adopt AI intelligently – distinguishing between when to use off-the-shelf tools and when to invest in deeper AI capabilities. His work spans architecture, strategy, and practical implementation, with a strong emphasis on critical thinking, rigor, and avoiding hype-driven decisions that waste millions.
Memorable Quotes
- “The person with the largest platform is not necessarily the person with the largest knowledge.”
- “What we have today are opinions largely, and very few factual information.”
- “There are no AI experts. Everybody is learning.”
- “AI is not a replacement for a practitioner – it’s a power tool to work alongside them.”
- “Good engineers won’t be replaced by AI. They’ll be turbocharged by it.”
- “Fast food has a market, but it hasn’t killed gourmet.”
- “We should not take an opinion as gospel just because another million people listen to it.”
- “What we see scares us. What we don’t, we don’t care. That’s human nature.”