"People will go to a website because they're going to enjoy the experience there, not because the information they need is there."

I talk to Chus, a Barcelona-based art director and UX/UI designer with 15+ years across Madrid ad agencies, an Amsterdam mobile startup, a gaming corporate, a trading platform, and now a Dutch insurer, about his side mission to make the internet weird again. By day he designs AI tools for insurance agents at Nationale-Nederlanden. By night he builds hyper-realistic folder icons and a portfolio that runs like a 90s operating system.

We get into why the Flash era was the best the web ever looked, how conversion obsession and design systems flattened everything, why he uses Lovable constantly but never ships the Lovable look, and his exact workflow from ChatGPT tech-stack advice to plan mode to telling the AI to stop adding emojis.

In this episode:

- Make the Internet Weird Again: Why AI answering your questions means websites now have to be worth visiting for the experience.

- The Flash Era: What 2004 got right, and how accessibility and conversion obsession turned the web flat and minimal.

- Bootstrap, Tailwind, Lovable: The same trap three times. Out-of-the-box defaults are why every AI site looks identical.

- Never Ship the Lovable Look: Using AI for the code you can't write, then replacing every visual default with your own.

- Agency vs Corporate: 15 years from late-night Madrid briefings to laser-focused insurance product teams, and why both can work.

- AI in Insurance: How a regulated Dutch insurer studies China's AI moves and builds agent tools without breaking the rules.

- The 140 Folders: From a post on X to a Lovable-built gallery and mini-shop, step by step.

- Intentional Design: A Comic Sans can work if it's on purpose. Limit the AI early, upload your own fonts and icons.

- Advice for Non-Designers: Learn the design movement behind what you call "nice" before you prompt anything.

🎧 Full episode on all podcast platforms

💬 Was the Flash-era internet actually better, or is it just nostalgia? Let us know in the comments!

🔔 Please like and subscribe! Every subscriber helps our channel grow.

#DN #WebDesign #Lovable #UXDesign #MakeTheInternetWeirdAgain #ArtDirector #AIDesign #VibeCoding #DesignSystems #Startups

Timestamps:

0:00 Intro

0:39 Madrid ad agencies: briefings, late nights, the golden rule

3:00 The jump to a mobile startup in Amsterdam

4:38 Learning how developers think without learning to code

6:00 Gaming, agencies, Barcelona, and the trading platform

7:00 Nacar design studio and the biggest corporate job yet

8:50 Make the internet weird again

9:55 Why the Flash era was the best the web ever looked

11:10 Nobody visits websites for information anymore

12:04 Design systems and why every website looks the same

13:31 Airbnb, Shopify, and the designers fighting back

14:45 Bootstrap, Lovable, and the out-of-the-box trap

16:12 Niklas's brokerage app: when a redesign kills the UX

18:58 Is this the NFT bubble again?

20:14 Local AI, open weights, and where the craft goes

22:43 Firefly, prompting, and why a designer now feels like a coder

26:35 AI inside a regulated insurer

29:21 The 140 folders: from an X post to a Lovable shop

32:50 Plan mode first, small prompts, stop the emojis

35:38 Advice for non-designers who want to build something nice

"I haven't met anybody who's working less after the release of AI. If anything, deadlines got shorter."

I talk to Alexander Solod, an AI developer who has been building with generative models since a GPT-2-written paper blew his mind in college, about where AI actually stands right now: open weight models running at home, on-device chips, the interface problem nobody has solved, and what AI is really doing to jobs.

We get into why Alex runs Gemma 4 on a Mac Studio for his personal agents, the "Airbnb for compute" idea of renting out your idle hardware for tokens, why radiologists are still booming ten years after AI was supposed to replace them, and the only career advice that survives automation: get into a position where you make decisions.

In this episode:

- Open Weights at Home: Why the future is small on-device models for daily work, with frontier models reserved for the hard stuff.

- The Benchmark Problem: Why "better than the average human at using a computer" tells you almost nothing.

- Filter and Enzyme: LLMs won't discover relativity, but they filter the world's information and collapse weeks of work into hours.

- Airbnb for Compute: Renting out your idle Mac Studio for tokens, and why solar at home makes it obvious.

- Chat Was V1: Voice in, visuals out, generative UI, and why the AI interface is still unsolved.

- Ambient AI Scribes: The easiest healthcare win so far, and why radiology still hasn't been automated.

- Nobody Works Less: The paradox of AI productivity, and why the layoffs blamed on AI are really pandemic overhiring.

- The Job That's Safe: A machine can never be held responsible, so get into a position with stakes and decisions.

🎧 Full episode on all podcast platforms

💬 Are you working less since AI, or more? Let us know in the comments!

🔔 Please like and subscribe! Every subscriber helps our channel grow.

#DN #OpenWeightModels #OnDeviceAI #AlexanderSolod #AIJobs #HealthcareAI #AIAgents #DecentralizedCompute #LocalLLM #Startups

Timestamps:

0:00 Intro

0:25 The GPT-2 paper that pulled Alex into AI

1:12 OpenAI Gym and the hide-and-seek video

3:28 Open weight models: running Gemma 4 on a Mac Studio

4:04 Nobody knows the limit of intelligence

4:53 Is "better than the average human" a real benchmark?

6:49 Could an LLM have discovered relativity?

8:39 LLMs as filter and enzyme

9:54 On-device AI and Apple's neural engine bet

11:36 Airbnb for compute: renting out your idle hardware

13:35 Chat was V1: the unsolved AI interface

15:43 How far away are brain interfaces?

16:52 Predictive healthcare and the radiology myth

19:28 Giving clinicians their time back

20:48 Regulation, trust, and security in healthcare AI

22:24 Ambient AI scribes: the easiest win so far

23:47 Whoop data and democratizing Bryan Johnson

26:05 Nobody is working less since AI

27:23 Are the layoffs really AI, or pandemic overhiring?

30:31 How to prepare: actually use the agents

32:08 Best advice Alex ever received: keep momentum

33:01 Hot take: the AI interface is still wide open

"People think TPM is gonna be the job that gets replaced first. In reality, it's gonna be the hardest thing to replace."

I talk to Jake Kim, founder of Serro AI, about agentic technical program management and why the horizontal coordination role everyone expects AI to kill first is actually the one it can't. Jake was an engineering manager at Apple, where his team's software sat on the critical path to every iPhone and iPad launch, and where he watched world-class TPMs hold the whole machine together.

We get into the temporal knowledge graphs that give AI agents organizational memory, the client with 2 TPMs, 300 people, and 75 recurring reports, the R&D tax documentation wedge that pays for the product on day one, and the founder lesson Jake learned right before running out of money.

In this episode:

- The Job AI Can't Replace: Why vertical agents (coding, design, product) make the horizontal coordinator more essential, not less.

- Agentic TPM: Technical program management, not manager. Extending human TPMs' reach to the agents downstream.

- Temporal Knowledge Graphs: Why programs need a memory of decisions, tradeoffs, and dependencies over time, not just a snapshot.

- 2 TPMs, 300 People, 75 Reports: The middle-management squeeze nobody's tooling has touched.

- The R&D Tax Wedge: The audit-ready documentation report worth millions in credits that gets Serro in the enterprise door.

- Per-Program Pricing: Why Serro charges $100 per tracked program with unlimited seats, and what that alignment does in enterprise sales.

- Don't Be Shy About Your Vision: The lesson Jake learned right before running out of money, and why humble framing nearly killed the company.

- Advice for TPMs: Why the role should feel confident, not threatened, and how to extend your influence to the agent layer.

🎧 Full episode on all podcast platforms

💬 Is middle management really the hardest thing for AI to replace, or the first to go? Let us know in the comments!

🔔 Please like and subscribe! Every subscriber helps our channel grow.

#DN #AgenticTPM #KnowledgeGraphs #JakeKim #SerroAI #AIAgents #EngineeringManagement #EnterpriseAI #Startups #TPM

Timestamps:

0:00 Intro

0:15 What agentic TPM actually is

1:37 From Apple engineering manager to founder

3:12 Why AI employees haven't worked so far

4:46 Why agentic TPM is technically hard (programs vs projects)

8:22 Integrating with the company's tools

9:21 2 TPMs, 300 people, 75 recurring reports

13:15 Temporal knowledge graphs explained

14:28 How the product works: Gantt charts, drift alerts, MCP

16:00 Two years of pivots and what they taught him

17:50 The vision lesson: don't be shy right before running out of money

19:44 The R&D tax documentation wedge (and PwC)

22:38 Per-program pricing and the death-of-SaaS question

25:58 Why not just be a knowledge graph company?

27:54 The AI-native org two to three years

"A good assistant takes about three days for one blueprint. I'm using AI, and it's 30 seconds."

I talk to Vinh, solo founder of HD Parking Lot, about the construction marketplace he's building after a $25,000 junk-removal quote pushed him into the most overlooked trillion-dollar industry in tech. Vinh spent 20+ years as a software engineer before a rough patch (losing his mom, getting laid off) turned into a junk-removal side hustle, which turned into discovering how broken the contractor-homeowner relationship actually is.

We get into why Angi and Thumbtack sell your information to 20 contractors and then abandon you, the AI blueprint reader that compresses three days of estimation work into 30 seconds, and why the boring industries nobody wants to touch are where the real opportunities hide.

In this episode:

- The $25K Junk Quote: The cleanup bill that turned a laid-off engineer into a junk removal operator, and then a construction founder.

- The Lead-Selling Trap: Why Angi and Thumbtack make homeowners repeat themselves to 20 contractors, and what a marketplace that doesn't sell leads looks like.

- The AI Blueprint Reader: Materials, labor hours, schedule, and regulations from an uploaded blueprint in 30 seconds instead of 3 days.

- Boring Industry Gold: Why construction is a trillion-dollar industry tech people won't touch, and why that's exactly the opportunity.

- The Full Lifecycle Bet: Directory, project bidding, materials marketplace, calculators, and community in one platform that stays past the handshake.

- The Visualizer: The free tool that repaints your room from a photo in 30 seconds, built to settle an argument with his wife.

- Don't Do Paper Research: Why one day living inside the process you want to automate beats a hundred customer interviews.

- Kitesurfing as Focus Training: Why you can't ride the wind while worrying about your startup, and what that teaches about adaptability.

🎧 Full episode on all podcast platforms

💬 Would you rather build in the crowded AI market, or own a boring trillion-dollar industry? Let us know in the comments!

🔔 Please like and subscribe! Every subscriber helps our channel grow.

#DN #ConstructionTech #AIBlueprints #Vinh #HDParkingLot #BoringIndustries #SoloFounder #BuildingInPublic #Startups #PropTech

Timestamps (RAW — re-time after edit):

0:00 Intro

0:35 From junk removal side hustle to construction tech

4:14 The lead-selling trap: how Angi and Thumbtack actually work

8:33 The $25K junk quote that started everything

11:32 Why boring industries beat flashy AI startups

13:03 The AI blueprint reader: 3 days to 30 seconds

14:40 The Visualizer: repaint your room from a photo

18:37 The full feature walkthrough: directory, bidding, materials, calculators

26:26 Why he builds in public (the Marc Lou inspiration)

30:16 Marketing as a solo founder with an SEO background

33:32 Building a global company with an equity-based team

44:14 The one advice: hands-on research, not paper research

46:51 Kitesurfing, focus, and adaptability

"Kids don't just consume technology, they create with it."

I talk to Feifei Qiu, founder of Kindlewood Studios, about the dragon story her four-year-old told at the dinner table that turned into a learning platform, and why storytelling is the wedge into teaching kids reading, writing, and creativity. Feifei spent years as a product manager at Microsoft before leaving to build Kindlewood for her own kids, and 200+ young storytellers later, the thesis is holding.

We get into the viral loop of imagination (one kid's story inspires five more), how she designs AI that keeps kids in the driver's seat instead of writing for them, and the product lessons she carried over from enterprise PM work.

In this episode:

- The Viral Loop of Imagination: How one dragon story at a play date turned every kid in the room into a storyteller.

- Kids as Creators: Why children are born to create, and what happens when you give them the tool to express it.

- AI in the Passenger Seat: How Kindlewood's writing coach asks open-ended questions instead of writing the story, keeping kids as the authors.

- The Persona Split: Why the reading app and the creator experience had to be completely separate products, a lesson from watching her three-year-old.

- From Microsoft PM to Founder: What enterprise product thinking transfers to a kids' startup, and what you have to unlearn.

- Demanding Customers Are the Best Customers: The PM lesson that the hardest customers to win become the most loyal.

- Steady Over Viral: Why she's deliberately growing slow, with 9 school partnerships and workshops where kids don't want to leave.

- "Is It Me the Reason You Got a Job?": The five-year-old's question that sums up the whole company.

🎧 Full episode on all podcast platforms

💬 Should kids use AI to create, or does it risk replacing their creativity? Let us know in the comments!

🔔 Please like and subscribe! Every subscriber helps our channel grow.

#DN #EdTech #KidsLearning #FeifeiQiu #Kindlewood #Storytelling #AIinEducation #Startups #Parenting #Creativity

Timestamps:

0:00 Intro

0:51 Why build for the youngest learners

2:30 From China to Seattle to Microsoft

3:55 Big tech vs founding: what actually changed

7:12 Product thinking she carried from Microsoft

11:29 What you have to unlearn after big tech

14:12 Why the name Kindlewood

16:16 The dragon story that started everything

21:04 Traction: 200+ young creators, 9 school partnerships

24:37 Ten years of overnight success

26:04 The viral loop of imagination

28:07 AI in education: kids stay in the driver's seat

31:36 How vibe coding changed her builder workflow

34:25 "Is it me the reason you got a job?"

"It's cheaper to upgrade to an expensive model."

I talk to Joakim William Hauge, co-founder of Monetize, about "runtime economics" (the new category he's carving out for AI governance) and why CFOs are locked out of the AI dashboards their engineers use every day. Joakim spent years in quantitative finance and risk management before turning that framework onto the non-deterministic runtime behavior of autonomous AI systems.

He's building Unitflow, an enterprise governance layer launching soon, and already shipped Circuit Breaker on GitHub as a validation vehicle. The wedge: use runtime telemetry to translate token spend into income-statement language the CFO can actually act on.

In this episode:

- Runtime Economics: The new category Joakim is defining, using runtime telemetry to understand AI value creation and risks, not just cost optimization.

- The CFO Gap: Why the executives signing off on AI budgets are locked out of the dashboards their engineers actually use.

- Why Expensive Models Are Cheaper: The counter-intuitive result of testing, that upgrading to a more expensive model can reduce total spend by killing recursive retry loops.

- Cost, Value, Risk: The three-pillar framework for governing AI spend, and why CFOs need income-statement language, not token dashboards.

- Circuit Breaker: The open-source runtime governance tool Joakim shipped to validate the thesis before Unitflow's full launch.

- The Uber Cost Shock: How cost-cutting headlines from big companies changed the tone of AI adoption in a matter of months.

- The Usage-Based Pricing Trap: Why real-time pricing for non-deterministic AI runtime is one of the hardest unsolved problems in SaaS.

- Understand the Customer's Story: Joakim's biggest founder lesson, that the shortcut isn't shipping fast, it's understanding what the buyer actually needs.

🎧 Full episode on all podcast platforms

💬 Should CFOs get their own AI dashboards, or is token spend just an engineering-team problem? Let us know in the comments!

🔔 Please like and subscribe! Every subscriber helps our channel grow.

#DN #RuntimeEconomics #AIGovernance #JoakimHauge #Unitflow #AICostManagement #Startups #AIAgents #EnterpriseAI #FinTech

Timestamps:

0:00 Intro

0:40 From quant finance and risk management to AI governance

2:19 Why he's building in public

5:32 The engineer vs CFO stakeholder gap in AI adoption

7:56 Runtime economics defined in one sentence

8:20 The Uber cost shock: why April changed everything

14:06 Volatility in AI cost and how to price it

19:34 What Unitflow actually does: the three pillars

23:54 Why CFOs are locked out of the AI dashboards

25:11 Frontier vs cheaper models: when to use each

27:09 Why expensive models can actually be cheaper

27:44 The GitHub Circuit Breaker tool

30:43 Founder lesson: understand the customer's story

34:04 The unsolved problem of real-time usage-based pricing

"Software will be removed. Agents are gonna be what you sell."

I talk to Julian Fitzpatrick, founder of Build with Pixel, about why packages of AI agents will replace B2B software for small businesses, and the bottleneck-and-propagate playbook he uses to build tools one company at a time. Julian spent 17 years in visual effects before pivoting into AI when ChatGPT 3 dropped, and now runs a two-pronged model: consulting local small businesses to automate their workflows, then turning those tools into a public platform.

We get into the volleyball academy case study, why he thinks the gap between frontier and open-weight models is about to disappear, the three-layer human-in-the-loop workflow for shipping agent code safely, and why he's building everything in public.

In this episode:

- Agents as the New Software: Why Julian thinks packages of agents will replace B2B software, tailored per vertical instead of per feature.

- The Bottleneck-and-Propagate Playbook: How he wins one small business (a volleyball academy, an architecture firm), automates a workflow, then propagates the tool across the vertical.

- Local Models Closing the Gap: Why open-weight models are approaching frontier quality, and what that means for on-prem SMB deployments.

- The Three-Layer Approval Workflow: Screenshots, staging, production. The human-in-the-loop pattern that lets agents ship code without breaking things.

- Building in Public as a Solo Founder: Why Julian shares everything he builds, and how that becomes trust-building distribution.

- The AI Operator Role: The new job category emerging as agents get real — humans as orchestrators, not executors.

- From VFX to AI Agents: 17 years in visual effects, then a pivot into agent infrastructure when ChatGPT 3 dropped.

- Signal vs Noise: The Kevin O'Leary / Steve Jobs framing Julian uses to stay focused as a solo builder.

🎧 Full episode on all podcast platforms

💬 Will "packages of agents" actually replace B2B SaaS, or just add a new layer on top of it? Let us know in the comments!

🔔 Please like and subscribe! Every subscriber helps our channel grow.

#DN #AIAgents #SmallBusinessAI #JulianFitzpatrick #BuildWithPixel #BuildingInPublic #Startups #SMB #AI #AgenticAI

Timestamps:

0:00 Intro

0:49 What Build with Pixel is and who it serves

2:00 From 17 years in VFX to building AI agents

3:28 Open-weight models closing the gap on frontier

5:12 The local-vs-cloud choice for small business AI

8:27 The bottleneck-and-propagate methodology

9:35 Case study: the volleyball academy CTA workflow

11:43 Which industries are easiest to enter with AI now

14:17 Why he's building in public

17:06 Attention is all you need, audience as the moat

22:49 Why he doesn't obsess about ICP or revenue ranges

25:49 Steve Jobs, signal vs noise, and the solo founder trap

26:20 The 12-month arc from prompts to harnesses to agents-as-software

30:19 Human-in-the-loop and the three-layer approval workflow

33:44 Where to find Julian and Build with Pixel

"Distribution is the new moat. Five years ago it was your product."

I talk to Calvin Lim, co-founder of Clarity AI, about building AI-native analytics for the coffee shop owners and family-business founders that every tech tool ignores, and why distribution has replaced product as the defensible layer for anyone building software today. Calvin grew up in San Francisco, ran a sneaker business through high school, did stints in electronics and Amazon retail, and most recently worked in analytics with sales teams before going full-time on Clarity.

We get into why AI SDRs are saturating B2B email (and why open rates don't mean anything anymore), why real-life events are the unreplicatable distribution channel everyone's missing, the open-weight token war, and the SF founder mindset Calvin uses to avoid getting distracted by the person next to him raising $25M.

In this episode:

- The New Moat: Why distribution has replaced product as the defensible layer, and what changes for founders building today.

- AI Analytics for SMBs: Why coffee shop owners and family hotels are the underserved customers Clarity AI is built for.

- The Three-Layer Stack: Analytics, intelligence, orchestration. Why most data tools stop at the dashboard.

- Why AI SDRs Don't Work: Why everyone using Clay and Apollo sources from the same list, and why email open rates don't mean anything anymore.

- Real Events Over Cold Outreach: Why grassroots and IRL events are the distribution channel that can't be copied.

- The SMB to Enterprise Playbook: Why starting with smaller customers (the Salesforce path) can still be the right move in the AI era.

- The Open-Weight Token War: Why Chinese AI models reselling tokens at 93% cheaper is the disruption nobody's pricing in.

- Stay Rooted in Substance: The SF founder mindset Calvin uses to avoid chasing the person next to him raising $25M.

🎧 Full episode on all podcast platforms

💬 Has email outreach actually died, or is it just that everyone's using the same tools? Let us know in the comments!

🔔 Please like and subscribe! Every subscriber helps our channel grow.

#DN #DistributionMoat #AIforSMB #CalvinLim #ClarityAI #SaaS #SanFrancisco #Startups #AIAgents #FoundersJourney

Timestamps:

0:00 Intro

0:44 What Clarity AI is and who it's for

3:25 How the platform works for a first-time user

4:33 The pivot from no-code dashboards to AI-native

6:02 Distribution is the new moat

9:14 Why selling small businesses needs human-to-human

12:10 AI SDRs and the spam problem in cold outreach

14:21 Why everyone using Clay and Apollo competes on the same list

15:41 The end of fat SaaS margins

17:04 China reselling Anthropic tokens 93% cheaper

19:19 The future of local AI compute

23:51 Calvin's path from sneakers to analytics

28:40 Why San Francisco is the center for builders

33:16 The Salesforce SMB to Enterprise playbook

35:32 What changes for SMB analytics in the next 12-24 months

"Our AI doesn't replace underwriters. It does the boring work so they can make the decisions."

I talk to Danny Tang, founder of Tradata, about why the non-qualified mortgage market is the underserved $239 billion slice of US mortgages, and what it takes to build AI for an industry that demands 100% accuracy. Danny grew up in Hong Kong, spent four years in the financial industry there, and moved to San Francisco to build mortgage tech that runs on-prem instead of in the cloud.

Tradata interviewed 60+ underwriters before writing a line of product code. The goal: take the 1+ hour per file underwriters spend reading 400-page bank statements and turn it into minutes, without ever leaking data outside the lender's environment.

In this episode:

- The $239B Non-QM Wedge: Why a $239 billion segment (9% of US mortgages today) is set to become 30%+ of the market over the next decade.

- The 100% Accuracy Bar: Why regulated lenders can't use general-purpose LLMs, and what it takes to clear that bar without buyback risk.

- On-Prem Over Cloud: Why Tradata runs locally on customer infrastructure, and why finance is moving in the opposite direction of every other industry.

- Document Hell: The 400-page bank statements, tax returns, and PNLs underwriters wade through, and the 1+ hour per file that disappears with AI.

- Why Now: Gig workers, creators, and entrepreneurs need homes but don't qualify for traditional mortgages, and the trend is only accelerating.

- The Lender Tech Gap: Why mortgage giants like A&D are building in-house AI while small and medium lenders fall behind, and the market consolidation that follows.

- AI Augments, Doesn't Replace: Why underwriters fear AI tools that promise to replace them, and what changes when the AI does the boring work instead.

- From Hong Kong Finance to SF Startup: Danny's path from the reinsurance group in Hong Kong to building mortgage tech in San Francisco.

🎧 Full episode on all podcast platforms

💬 Should regulated industries stay on-prem, or move to cloud-native AI? Let us know in the comments!

🔔 Please like and subscribe! Every subscriber helps our channel grow.

#DN #Mortgage #FinTech #AIinFinance #DannyTang #Tradata #Startups #NonQM #PrivateAI #SaaS

Timestamps:

0:00 Intro

0:18 From Hong Kong finance to founding Tradata in San Francisco

2:30 The document-processing pain in non-QM underwriting

4:40 The $239B non-QM market and why it's growing

7:00 Building for 100% accuracy in a regulated industry

7:44 Private on-prem AI vs cloud-native software

9:30 The underwriter workflow Tradata replaces

12:00 Why non-QM specifically (gig workers, creators, real estate)

15:00 Self-teaching code on document processing side projects

20:00 The lender tech gap: giants vs small lenders

24:00 Solving problems no one cares about

27:30 Growing up on Rich Dad Poor Dad, Ray Dalio, Warren Buffett

"Two people reading the same book get completely different action steps."

I talk to Shruta Satam, co-founder of Pustak, about the knowing-doing gap — why most of what we read never makes it into our lives — and how Pustak personalizes action steps from every book to the reader's actual goals. Shruta spent 18+ years in risk management consulting at Deloitte and PwC across the US and Australia before leaving to build Pustak with her technical co-founder.

We get into why summaries alone aren't enough, what the Pustak Coach actually does, the localization push for India, and the marketing realization most first-time founders learn the hard way.

In this episode:

- The Knowing-Doing Gap: Why less than 10% of what you read translates into actual behavior change.

- Personalized Action Steps: How two readers of the same book get completely different recommendations from Pustak's engine.

- Beyond Book Summaries: Why most reading apps stop at summarizing, and what closing the loop actually requires.

- The Pustak Coach: The habit tracker that turns "I read this" into "I'm doing this."

- 1,000+ Curated Books: Why she chose hand-picked esoteric books over bestseller volume.

- Localization for India: The market entry strategy and the regional-language translation push.

- From Consulting to Startup: The risk frameworks that translate, and the ones you have to leave behind.

- The Marketing Surprise: Why great products often sit unknown, and the lesson every first-time founder hits.

🎧 Full episode on all podcast platforms

💬 Do book summary apps actually move the needle on behavior change, or do they just feel productive? Let us know in the comments!

🔔 Please like and subscribe! Every subscriber helps our channel grow.

#DN #Pustak #SelfHelp #ProductivityApps #ShrutaSatam #StartupJourney #ContinuousLearning #AIPersonalization #BookSummaries #Startups

Timestamps:

0:00 Intro

0:37 From 18 years at Deloitte and PwC to founding Pustak

2:57 The knowing-doing gap (less than 10% applied)

5:32 Personalization and per-reader action steps

7:04 Why retention is the biggest hurdle

8:18 How AI powers customization for each user

11:17 Working with her technical co-founder

12:39 The future of AI-powered learning

14:31 The roadmap: app launch and India localization

15:49 Competitors: Blinkist, Headway, and what's different

17:15 Curated books vs reading in full

22:30 Startup risk vs corporate risk management

24:26 What consulting teaches you about building products

30:50 The Pustak Coach habit tracker in detail

34:10 The marketing realization most founders hit