About This Episode

Bhairav sits down with Sanjay Rakshit, now at Poppulo, for his third or fourth appearance on the show.

The conversation was sparked by a quote Sanjay gave at DSE (the digital signage industry event run by Invidis): "LLMs are the most overrated part of AI." They unpack what that actually means, and dig into agents, agentic systems, context engineering, data fidelity, AGI, robotics, and the future of engineering jobs.

What We Cover

- What Sanjay meant by his "LLMs are overrated" quote, and the DSE event where he said it

- The difference between an agent and a chatbot: agency, planning, and achieving an outcome versus reactive output

- Where LLMs actually fit inside an agentic system (as one tool among many)

- Why context engineering, not the LLM call, is where most of the real work happens

- Data fidelity versus data quality, and why it matters more than raw data volume

- The mathematics underpinning AI, and why understanding first principles beats following influencers

- Quantum computing's potential impact on AI going forward

- AI, jobs, and the "power tools" analogy for engineering productivity

- Sanjay's take on AGI, and why he remains sceptical

- Robotics versus AGI, and his work with the Bayes Institute in Scotland

Key Takeaways

- An agent is any entity that can think, plan, and act to achieve an outcome; a chatbot just responds

- Roughly 70% of agentic system work is context generation and systems engineering, not the LLM call itself

- Data fidelity (relevance to the use case) matters more than raw data quality

- AGI scepticism rests on the argument that the brain's capacity may be too vast to model

- AI is compressing the mechanical effort of coding, freeing engineers to focus on solving the actual problem

Links

Connect with Sanjay Rakshit on LinkedIn: https://www.linkedin.com/in/sanjay-rakshit/

Sanjay is back! But this time he answers the famous five questions...

https://www.linkedin.com/in/sanjay-rakshit/

Guest: Mateo Bervejillo, CEO of Xseed and author of Back to Basics

Mateo started his career as an attorney in Uruguay before moving into outsourcing operations, growing into VP-level roles leading teams across the US, Uruguay and India, and eventually becoming CEO of Xseed, a LATAM software talent company he helped grow from five people to a hundred. He has spent two decades in the gap between being good at a job and being good at leading people who do that job.

What we cover in this episode

The conversation covers what actually happens when you get promoted into leadership for the first time, and why the skills that earned you the role are rarely the ones the role requires. We talk about inheriting teams versus building them, the mistake of leading by authority rather than trust, why scaling a business means changing how you work rather than just working more, and how to tell whether you are actually succeeding as a leader.

Key topics and timestamps

[00:02] Why a legal background can become a "business prevention syndrome" and how to get past it

[08:00] The transition from corporate to entrepreneurship, and losing the fear of rejection

[11:00] Building a network before you need it: your 20s, 30s and 40s

[14:45] The book: what nobody tells you when they give you a team

[19:50] Inheriting a team, taking it slowly, and giving people the right chance

[22:30] Leading by authority versus leading by alignment

[26:45] Building a team from scratch, cultural fit, and hiring for the stage you are actually at

[29:00] Why unconscious bias in hiring is unavoidable, and how to design around it

[33:50] Scaling and the player-coach problem: when letting go of the work feels like not working

[41:00] Leadership myths: you are on camera constantly, and the little things matter more than you think

[45:50] How to know if it is actually working: results, attrition and growth

[48:00] Advice to his younger self: network outside the company, and start early

Mentioned in this episode

Back to Basics: Bridging the Gap Between Expectations and Preparation for the Leader Role by Mateo Bervejillo, available on Amazon in Kindle, paperback and hardcover. Also at www.xseed.com.uy/back-to-basics-book

First Break All the Rules by Gallup (on employee motivation and engagement)

Connect with Mateo

Company: www.xseed.com.uy

LinkedIn: search Mateo Bervejillo

Email: [email protected]

The Atom CTO Podcast is hosted by Bhairav Patel. New episodes cover technology, entrepreneurship and the people building things that matter. Subscribe wherever you listen to podcasts.

This week's guest is Mateo Bervejilo - CEO of Exceed Solutions, who will be answering the famous 5 questions

# Show Notes: Vibe Coding and Real World Problems: Why AI Still Needs a CTO

**Guest:** Anuraag Jain, CTO, Executive Coach and Author

**Website:** [growthvariable.com](https://www.growthvariable.com)

**LinkedIn:** [Anuraag Jain](https://www.linkedin.com/in/anuraagjain)

---

## About This Episode

Anuraag Jain is a 30-year tech industry veteran who started as an Oracle DBA in India without an engineering degree and became a CTO in under nine years. He was the first developer hired at Betfair, where he built the betting exchange engine behind one of the era's biggest fintech IPOs, and served as CTO at The Hut Group (THG) during their run as the UK's fastest-growing tech company.

In this episode, Bhairav and Anuraag talk honestly about vibe coding: what it can do, where it breaks down, and why thirty years of building things under pressure still matters in an AI-assisted world.

---

## What We Cover

- How two non-traditional hires found their way into tech and why that background shaped how they build

- The early days at Betfair: manual bet settlement, shoestring infrastructure, and recovering 36 hours of betting data from server log files at 2am

- What scaling really means when you have no safety net

- The 80/20 rule for vibe coding: what AI handles well and what still requires human judgment

- Security blind spots in AI-generated code and why even experienced engineers get caught out

- The hidden cost of running AI tooling against a large code base

- Repo to Pitch: the open source tool that turns a code base into a full pitch deck and documentation suite

- RepoWatch: how to track exactly what AI has changed in your code, commit by commit

- *Prompt Engineering the Subconscious*: the book that connects AI prompting to mindset and reached the UK top 50

- What the fractional CTO market actually looks like right now, and the three types operating in it

- Practical advice for anyone trying to break into tech or move toward a CTO role today

---

## Key Takeaways

- AI can handle roughly 80% of what used to require a developer. The remaining 20% is the domain logic that requires someone who actually understands the problem.

- The AI does not know which parts of your code cannot be touched. Unless you tell it, it will modify anything, and you may not notice until something breaks.

- Vibe coding a product does not make it production-ready. Security, scalability, and edge-case logic still need human attention.

- Building something real, however small, is still the best way to develop the judgment the CTO role requires. The tooling available now means there is no excuse not to.

---

## Links

- Anuraag Jain on LinkedIn: [linkedin.com/in/anuraagjain](https://www.linkedin.com/in/anuraagjain)

- Growth Variable: [growthvariable.com](https://www.growthvariable.com)

- Repo to Pitch (open source): [growthvariable.com/repotopitch.html](https://www.growthvariable.com/repotopitch.html)

- RepoWatch: [growthvariable.com](https://www.growthvariable.com)

- *Prompt Engineering the Subconscious*: [book.growthvariable.com](https://book.growthvariable.com)

---

## About the Atom CTO Podcast

The Atom CTO Podcast, transitioning to Bits, Bytes and Business, features conversations with founders, CTOs and entrepreneurs about building technology businesses in the real world. Hosted by Bhairav Patel.

Subscribe wherever you get your podcasts.

This week veteran tech CTO Anuraag Jain, author of Prompt Engineering the Subconscious is answering 5 questions: https://book.growthvariable.com/ i

James found me online looking for fractional CTO services. That conversation eventually made me an investor. In this episode I sit down with all three co-founders of Koloso — James, Petra and Sowi — to talk about building a meaningful EdTech product in Zambia with limited capital, no technical background, and a problem that affects millions of children across Sub-Saharan Africa.

**The problem:** Zambian classrooms regularly have 60 to 100 children and one teacher. Teachers have no reliable way to track who is keeping up or falling behind, and a mountain of admin on top. The slogans around leaving no child behind simply don't translate into classroom reality.

**The product:** Children play a two-minute, ten-question quiz aligned to the Zambian national curriculum. Teachers get a real-time dashboard showing individual progress over time. Parents receive practical tips for reinforcing learning at home, in everyday Zambian contexts. Version two covers the entire teaching loop, from lesson planning through to compliance reporting.

**The challenges:** Schools are cautious adopters. The "lazy teacher" objection was real and unexpected. Annual procurement cycles create brutal seasonal gaps. Building without technical founders nearly finished them.

**AI at the right moment:** Koloso started using AI-assisted development around eight to ten months before this conversation. Problems that once required lengthy specifications now get resolved in half an hour. But if AI had arrived two years earlier they would have built the wrong product. A year later and they would have run out of money. The timing was critical.

**An African solution to an African problem:** Khan Academy hasn't gained traction in Zambia because it isn't aligned to the Zambian curriculum. Koloso's questions use Zambian names, Zambian currency, and Zambian contexts. The family-centred learning philosophy reflects Zambian cultural values in a way that imported solutions don't.

---

## Timestamps

| Time | Topic |

|------|-------|

| 00:32 | How the founders met and their backgrounds |

| 04:00 | Why data is the core problem in Zambian classrooms |

| 07:00 | Building the first prototype |

| 11:10 | How the app works |

| 13:58 | School adoption challenges |

| 17:28 | African solution to an African problem |

| 26:53 | AI-assisted development and why timing mattered |

| 35:01 | Advice to their younger selves |

| 37:21 | Where to find Koloso |

---

Koloso is live in Zambia, South Africa, Nigeria and Uganda, with Malawi in progress. Available on iOS and Google Play. Pre-seed round of $250,000 currently open. Actively seeking investors and schools.

**www.koloso.app**

*Subscribe to the Atom CTO Podcast for more conversations with founders building things that matter.*

This week I'm joined by Sowi, Petra and James, the founders of Koloso, a Zambian EdTech Startup

We're back! A new season of the Atom CTO podcast will be coming to your ears over the coming weeks so stay tuned!

What used to be relatively straightforward—“Twitter for debate, Facebook for fun, Instagram for cool visuals, LinkedIn for business”—has become chaotic, politicized, and in many ways more dangerous for founders and brands.

In this episode with Colin Kelly we explore:

-Why the old rules of social media are gone and why nobody can honestly claim to “know” how it all works anymore.

-How social platforms have shifted from organic discovery engines to paid advertising systems and what that means for startups.

-The renewed importance of old‑school sales and marketing (phone calls, events, letters) and using social media as supporting evidence rather than your primary growth engine.

-The tension around brand values, politics, and polarisation, and why being loudly “values‑driven” online can now backfire.

-How to show your human side online without falling into inauthentic “I’m crushing it” performance.

-Where AI, SEO, and search are heading, including a fascinating discovery about Instagram posts appearing in Google’s AI Overviews.