BIG IDEAS BY NEW ECONOMIES

Subscribe to stay ahead of technology trends. Never miss future editions.
Marco Zappacosta, co-founder and CEO of Thumbtack, joins NEW ECONOMIES to explain why Google search volume has reaccelerated to an all-time high post-ChatGPT, why the customers now converting through AI chat are more qualified and further into their decision than any channel Thumbtack has seen before, and why marketplaces for hiring humans, unlike commodity marketplaces such as Uber or food delivery, are mediated by certainty and confidence rather than speed and price.
About Thumbtack
Thumbtack is a technology company helping millions of people confidently care for and improve their homes. Every day in every county of the U.S., people turn to Thumbtack to complete small fixes, routine maintenance, and major improvements. With over 12 million 5-star projects and counting, they help homeowners and home professionals accomplish more.
Watch Now: Marco Zappacosta - co-founder of Thumbtack
We also cover why word of mouth, not a competitor, is Thumbtack’s biggest threat, capturing 80% of home-services demand through calls to neighbors and posts in group chats, and how LLMs are finally solving a personalization problem Thumbtack couldn’t crack in 20 years, replacing one generic question set per category with fully bespoke, project-specific questioning, using Marco’s own Murphy bed installation as the test case. We get into why almost no startup has survived in a marketplace category with close to a trillion dollars of spend, and what half the Thumbtack product team is now rebuilding around AI.
We close on the interface shift Marco almost missed, voice, not text, the board seat he still wants to fill, and where he expects human-capital marketplaces to go over the next 20 years.
This was a fascinating episode! Available everywhere you listen to podcasts.
Watch or listen now across YouTube, Apple Podcasts, Spotify, and X
Download the transcript 👇
Timestamps
(0:00) Meet Marco Zappacosta(1:57) Thumbtack Turns Nearly 20(3:00) Why Marketplaces Are Challenging(6:30) Thumbtack's First 12 Months(8:57) How Thumbtack Uses AI Today(17:34) The Ideal Customer Profile(19:18) How Homeownership Is Changing(24:09) Why Experts Have High Expectations(26:55) Building Trust With Users in Today's Environment(29:03) Experts Communicating Offline(30:54) Tensions With Marketplaces(33:12) Integrating AI Into Thumbtack's Complex Stack(38:42) Where and How to Place Bets(42:04) The Next Big Opportunity(44:29) How Marco Runs Thumbtack(51:02) What Is Still Yet to Be Achieved?(52:14) Rapid Fire
Our notes from this conversation
1. AI customers convert better than search customers ever did.
Post-ChatGPT, Google search volume for Thumbtack-relevant categories has reaccelerated to its highest point ever, but the more telling shift is on the AI side: users arriving via ChatGPT or Claude are more qualified, more motivated, and further into the decision than a typical search customer. Volume is still low, but Marco is treating it as the leading indicator for how discovery gets rebuilt.
2. Word of mouth, not a competitor, is Thumbtack’s real adversary.
80% of home-services demand still flows through a call to a neighbor or a post in a group chat. Marco sees the AI moment as the first real chance to intercept that demand before it disappears into an informal network Thumbtack can’t see or monetize.
3. Hiring a human is not a commodity purchase.
Marketplaces like Uber and food delivery compete on speed and price because the average basket is under $50. Home services average around $1,000 per purchase with real consequences for getting it wrong, so the decision is mediated by certainty and peace of mind, not convenience.
4. Marketplaces are brutal to bootstrap — and that difficulty is the moat.
Despite near-trillion-dollar category spend, almost no home-services startup has survived alongside incumbents like Angi and Yelp. The same friction that kills most entrants is what protects the few that break through to compounding scale.
5. LLMs are solving a 20-year personalization problem overnight.
Thumbtack historically applied one generic question set per project category, fine for common jobs, useless for anything niche. Marco’s own Murphy bed installation became the test case: an LLM asked the right follow-up questions instantly, something no static form could match at that level of specificity.
6. Half the product team is now rebuilding around AI.
Marco describes the integration as touching everything, core matching, the customer and pro experience, pricing, refunds, monetization. It’s a ground-up rebuild, not a feature bolted on top.
7. Voice is the interface shift he almost missed.
Asked what he’s changed his mind on in the past year, Marco points to voice as input and output, not because it’s novel, but because it removes typing entirely for how his kids and Thumbtack’s pros interact with technology. He doesn’t think it kills the keyboard, but expects it to sit alongside it as a default mode.
The board seat he still wants to fill: a technologist.
His current board covers CFO, CEO, and COO backgrounds, but he’s missing a product-obsessed technologist, someone in the mold of Snap’s Evan Spiegel, to pressure-test where AI takes the product next.
Links
Follow Ollie on X - https://x.com/ollieforsyth.
Follow Marco on X - https://x.com/mlz.
Visit Thumbtack - http://thumbtack.com.
Partnership: Harmonic is the go-to startup database - https://harmonic.ai.
Partnership: Hostinger is as a go-to tool for builders. Subscribers receive 10% off here - https://hostinger.com/neweconomies.
Previous episodes include
See all previous episode here 👉
If you enjoyed this episode, help sustain our work by clicking ❤️ and 🔄 at the top of this post.



Get full access to NEW ECONOMIES at www.neweconomies.co/subscribe

Subscribe to stay ahead of technology trends. Never miss future editions.
Tanay Kothari, founder of Wispr Flow, joins NEW ECONOMIES to unpack why most startups don’t die from bad ideas but from chasing too many good ones, and why matching your number of initiatives to your organizational capacity matters more than simply working longer hours.
About Wispr Flow
Wispr Flow turns voice into clean text. 4x faster than typing. 85% zero-edit. Speak naturally. Write perfectly.
Watch now: Tanay Kothari - Founder of Wispr Flow
We also get into the future of voice dictation and whether the keyboard is on its way out; how Wispr built an enterprise sales motion from scratch in less than a year, now accounting for a third of the company's revenue and serving more than half of the Fortune 500; the pricing psychology behind why ChatGPT feels free while Claude feels like a paid product; and why Wispr chose to build its own voice models in-house instead of relying on third-party AI providers.
We close on the story behind Wispr’s tuk-tuk campaign in India, how Tanay structures his week acting as chief of staff to the whole company, and a quickfire round covering his dream board pick and a free idea for Wispr Flow for creators by Ollie.
Watch or listen now across YouTube, Apple Podcasts, Spotify, and X
Download the transcript 👇
Timestamps
(0:00) Meet Tanay Kothari(1:50) Why Tanay Started Wispr(4:12) Building a Rocket Ship with Discipline(8:52) Why Now Is the Moment for Voice Dictation(12:27) Launching Wispr in India(14:51) Why Computing Is Still So Expensive(17:48) Why You Must Listen to Customers(24:00) How Tanay Stays Focused(27:00) Wispr's Internal Product Roadmap(29:15) Ollie Becomes Tanay's Chief of Staff(32:58) How to Keep the Talent Bar High(37:10) Rapid-Fire Round
Our notes from this conversation
* Most startups don’t fail from bad ideas, they fail from mismatched capacities.
Tanay’s reframe: the opposite of distraction isn’t focus, because focus can point at the wrong thing just as easily as the right one. A thousand-person company can be laser-focused and still fail if what it’s focused on doesn’t match what it can actually execute. His fix is mechanical, match your number of initiatives to your organizational capacity, because a task that took five people to build takes ten or twenty to maintain.
* Enterprise wasn’t bolted on. It was built from zero in under a year.
Wispr went from a pure consumer motion to a third of total revenue coming from B2B, with 15,000 companies and more than half the Fortune 500 as customers including: Microsoft, Nvidia, Notion, Clay, and Klarna. Companies like Slack and Notion took four to eight years to make that same consumer-to-enterprise jump. Tanay’s team compressed it into twelve months by treating it as a different product, not a repackaged one.
* ChatGPT feels free. Claude feels paid. That gap is the whole game.
This isn’t a throwaway comparison, it’s Tanay’s actual pricing philosophy. Perception of price is one of the most under-used levers founders have, more powerful than the marketing budget behind it. It’s why ChatGPT crossed a billion monthly active users while Anthropic, by his account, remains far behind on volume despite the stronger product.
* When nobody had a good enough model, Wispr stopped shopping and started building.
The team tried routing through other providers first and found the accuracy ceiling too low for voice specifically. So they built their own frontier voice lab from scratch, 15 people today, headed to 40 by the end of the year. The bet is that harness engineering, squeezing frontier performance out of cheaper models, only gets you so far before you have to own the stack.
* The keyboard isn’t dying of old age. It’s being made obsolete by 700 million people who never wanted it.
Tanay’s most pointed number: 700 million people worldwide have dyslexia or a speech impediment, and for them typing isn’t friction, it’s the single worst way to interact with technology. Voice isn’t a UX preference for that group, it’s the first real unlock they’ve had.
* There’s no fixed job at the top, just whichever fire is biggest that week.
Tanay describes his own role as deliberately fluid: one week he’s PMing a launch, the next he’s deep in Figma five days a week, the one after that he’s building out a CRO’s B2B function from scratch. The throughline isn’t a job title, it’s finding “the most important dumpster fire in the company that is not being taken care of” and sitting in it until it’s solved.
* Five to ten hours a week with users isn’t research. It’s the entire strategy.
Tanay doesn’t outsource customer insight to a feedback form. He sits beside users, watches their day, and treats that time as non-negotiable, not because it’s good practice, but because he thinks no company has ever succeeded without it. The line he keeps coming back to: your single job is to figure out what people want and give it to them, and everything else is downstream of that.
Links
Follow Ollie on X - https://x.com/ollieforsythFollow Tanay on X - https://x.com/tankotsTry Wispr Flow - https://wisprflow.ai
Previous episodes include
If you enjoyed this episode, help sustain our work by clicking ❤️ and 🔄 at the top of this post.


Get full access to NEW ECONOMIES at www.neweconomies.co/subscribe

Subscribe to stay ahead of technology trends. Never miss future editions.
Cameron Adams, co-founder and Chief Product Officer at Canva, joins NEW ECONOMIES to explain why AI gives creators more options than ever but not more chances to send the right message, and why taste becomes the real opportunity once every tool produces passable designs.
Watch now: Cameron Adams - Co-Founder at Canva
We also get into the contrarian bet behind Canva’s early success - ignoring The Lean Startup playbook to spend six extra months on user testing before launch, plus where AI is actually moving inside the product, from Canva AI 2.0 to magic layers to voice as the next creative interface.
We close on the fourteen-year journey of co-founding Canva with Melanie Perkins and Cliff Obrecht, and what Cam thinks creators should be focusing on going forward.
Watch or listen now on YouTube, Apple Podcasts, Spotify, and X
Download the transcript
Timestamps
(00:00) Meet Cameron Adams(02:06) Canva Create 2.0(05:13) The Canva Founding Story(07:45) Canva's First 6–12 Months(09:45) Building Fanatical Early Users(11:37) Landing the First Users(15:35) How Canva Stays Relevant(17:03) Canva's Focus on AI(19:50) Taste Is the New Differentiator(21:46) What Design Platforms Do Best Today(25:24) Canva's Most Popular Features(27:04) Inside Canva's Product Roadmap(34:05) What Creators Should Focus On(36:35) What Cam Is Most Excited About(38:25) AI-Generated Content(42:22) Why Music Is the Next Big Trend(46:10) Anyone Can Build Now(49:12) Cam's AI Stack(50:34) Ollie Joins as Cam's Chief of Staff(51:28) Co-Founder Relationships(56:00) What's Next for Canva
Our notes from this conversation
* AI gives you more options, not more chances.
You can now generate ten, twenty, thirty plus versions of anything using these AI tools, but the audience still only lets you send one message, maybe two at most. Cam’s take: the volume of options AI produces doesn’t lower the stakes of choosing correctly, it raises them. Great taste matters more than ever.
* The moat isn’t the model, it’s the product workflow around it.
One prompt box spitting out one image is table stakes now - anyone can do it. Canva’s edge comes from pulling teams, brand context, and every stage of a project into a single loop, which is why a quarter of a billion people bring their colleagues, friends, and family into the suite of product offerings with them.
* Ignoring conventional wisdom was the actual growth hack.
Investors were pushing Canva to ship fast per the famous playbook: The Lean Startup by Eric Ries. They spent six extra months on user testing instead, betting that a polished first experience would turn users into fanatical fans. Cam believes that patience became Canva’s early advantage. By delaying launch until the experience felt polished, the team laid the foundations for Canva’s organic growth loop.
* Voice is the next interface opportunity.
Cam points to Africa as a voice-first market by necessity, places where typing was never the default way people interacted with technology. Whoever wins on voice interfaces wins access to users that keyboard-first products never reached. The death of the keyboard may be looming.
* Fully AI-generated content plateaus. AI-assisted creators don’t.
AI Micro-dramas out of China are going viral on the strength of human storytelling, not the fact that they’re AI-made. Cam’s bet: pure AI-generated content is a novelty that settles into a niche, while creators who use AI to extend their own taste keep compounding.
* Velocity and quality aren’t in tension anymore, they’re the same discipline.
Canva mapped every feature shipped over the previous three months and found it had delivered more product output than at any point in the company’s fourteen-year history. Cam’s explanation isn’t “we cut corners” - it’s that better internal tools let designers and engineers prototype more ideas and still nail the one that ships.
* Fourteen-year co-founder relationships survive on self-awareness, as well as chemistry.
Cam’s answer to what makes Canva’s founding team last isn’t just the shared vision, it’s each person knowing precisely what they’re great at, what they’re not, and staying a well-rounded contributor instead of hiding in one fixed lane. That’s what let three people cover for each other for over a decade without competing for the same territory.
Links
Follow Ollie on X: https://x.com/ollieforsyth
Follow Cameron on X: https://x.com/themaninblue
Sign up to Canva: https://www.canva.com
Listen to all previous episodes: https://www.neweconomies.co/podcast
Subscribe to Cam's newsletter: https://promptedwithcam.substack.com/
Subscribe to Cam's podcast: https://www.youtube.com/playlist?list=PLATYfhN6gQz_ynkjnu_d63u1Qp600SlBH
Previous episodes include
If you enjoyed this episode, help sustain our work by clicking ❤️ and 🔄 at the top of this post.
…..
Brought to you by Harmonic - The complete startup database.


Get full access to NEW ECONOMIES at www.neweconomies.co/subscribe

Subscribe to stay ahead of technology trends. Never miss future editions.
Consumers care less about how something was made.
Justine Moore is a Partner at Andreessen Horowitz investing across AI and consumer, and one of the earliest backers of ElevenLabs. In this episode, she breaks down why AI microdramas are becoming one of the fastest-growing entertainment formats in the world, why China’s microdrama market has already overtaken its domestic box office, and why the U.S is only now catching up.
Watch now: AI Microdramas Are Exploding
We also explore why the first wave of AI video creators were attention seekers rather than storytellers, why that’s changing fast as real creatives move into the space, and why Justine thinks the “AI slop” debate misses the point entirely: slop existed long before AI, and the label won’t matter once most content is partially AI-made anyway.
We close the episode on why Justine believes agents that work before you ask are the next real unlock in consumer AI.
Watch or listen now on YouTube, Apple Podcasts, Spotify, and X
Download the transcript
Timestamps
(0:00) Meet Justine Moore (1:52) Generative Media's Inflection Point (4:23) AI Microdramas Are Exploding (10:02) Why AI Dramas Are New Forms of Entertainment (13:52) How to Create AI Microdramas (17:28) The Adoption of AI Microdramas (22:00) Content Becoming Timely vs. Timeless (25:00) Should Creators Be Disclosing AI Features? (33:45) How to Build AI Generative Media Startups(40:13) Justine's Favorite Agents (41:45) Is Consumer Tech Back? (45:15) Justine's Startup Ideas (46:40) Founders to Watch
Our notes from this conversation
1. AI video is finally good enough to stand on its own.
Early AI videos attracted attention because they were novel. Today, that's no longer the story. As Justine describes, model quality has improved to the point where AI-generated video can hold a viewer's attention with a coherent storyline. The competitive advantage is shifting from the technology itself to the creativity of the people using it.
2. Cheaper AI production is opening opportunities that global studios will eventually adopt.
Microdramas are the clearest signals of what is actually possible. Creators are already using AI to generate backgrounds, visual effects and techniques that were previously handled with CGI. The expectation is that major film studios will follow the same path, not to replace production or talent, but to cut costs on specific parts of production.
3. Ollie started producing a micro drama. This is what is possible!
Ollie built a short microdrama around the Nike origin story to stress-test the opportunities directly. The narrative, design and aesthetics came together fast and relatively cheaply. There are still gaps where these platforms can improve, for example: transitions between chapters aren’t natural yet, storytelling in the creator’s tone of voice has a way to go until perfect, and the cost to create these compounds very quickly once you’re iterating and constantly editing.
Watch here

4. Consumers care less about how something was made
Justine’s instinct is that mass-market audiences aren’t selecting for or against something because it’s AI-made, they’re asking whether it’s good. The people fixated on provenance are concentrated on X and Reddit, which is a different audience than the one actually consuming the content at scale.
5. Timely vs. timeless is a more useful lens than AI vs. human.
The human vs. AI comparison is the wrong split. Slop predates AI entirely, it was never a tooling problem, it’s a quality-and-intent problem, which is why “timeless” content (built to hold up regardless of when or how it was made) survives that axis and disposable content doesn’t.
During the episode, we also talked about if creators should be disclosing if AI tools were used and if so how. However, Justine thinks labeling is a losing battle: content is heading toward being partially AI-made by default, at which point a label stops signaling anything useful.
6. The hardest part of building here right now is differentiation.
Competing with OpenAI or Google at the foundation-model layer has gotten expensive enough that most new entrants shouldn’t attempt it. The real contest is one layer up, at the app and workflow level: the question isn’t whether you can build on top of the models, it’s who you serve and why they stay instead of switching to the next thin wrapper.
7. AI agents will matter more for individual creators than for big companies.
Justine’s case is that solo creators and small teams are the most resource-constrained group in the market, so offloading admin and logistics to agents is a bigger unlock for them than for anyone already running a team to handle it.
Links
Follow Ollie on X - https://x.com/ollieforsyth
Follow Justine on X - https://x.com/venturetwins
Justine's market map on AI Microdramas:
Previous episodes include
If you enjoyed this episode, help sustain our work by clicking ❤️ and 🔄 at the top of this post.
…
Brought to you by Hostinger. Use code NEWECONOMIES for 10% off.


Get full access to NEW ECONOMIES at www.neweconomies.co/subscribe

Subscribe to stay ahead of technology trends. Never miss future editions.
Maor Shlomo, founder of Base44, and Mark Tluszcz, Chairman at Wix, join NEW ECONOMIES to unpack the fastest-moving acquisition story in vibe coding: a solo founder who hit $1M in annualized revenue three weeks after launch, sold Base44 to Wix for $80M within months, and grew the business from roughly $3M to $150M in annualized revenue in the year since the deal closed.
Maor walks through the origin story — building a tool for his now-wife’s CRM problem, wanting to give non-technical people a way to build tools rather than just websites — and the operational chaos that followed, including running the platform solo with no monitoring and fielding a security scare in the middle of his brother’s wedding.
Mark also explains why Wix, twenty years into believing non-technical people deserve real building tools, saw Base44 not as a bolt-on but as the natural extension of what Wix had been doing since 2006.
WATCH NOW: $1M ARR IN 3 WEEKS
Throughout the episode, we get into why owning the full stack — front end, back end, and now the model layer — was non-negotiable for both companies, culminating in the release of Base 1, Base44’s own fine-tuned model for building web applications, and why Maor thinks the real bottleneck has already shifted from writing code to making good product decisions: as coding stops being the constraint, taste, distribution, and the judgment to know what’s worth building become the differentiators.
We also cover the coming wave of vibe-coded enterprise tools, why “verticalized” models beat horizontal ones for this category, and the emerging idea of Base44 acting less like a coding assistant and more like a co-founder.
We close on where Maor and Mark think the next big opportunities sit — re-humanizing technology through live events and the “off planet opportunity” (space) opened up by the SpaceX IPO — and who each of them would want on their board, landing, unexpectedly, on…
Watch or listen now on YouTube, Apple Podcasts, Spotify, and X
Download the transcript
Timestamps
(0:00) Meet Maor Shlomo & Mark Tluszcz(2:00) The Origin of Wix(4:07) The Base44 Founding Story(7:30) $1M in Revenue in Just 3 Weeks(9:23) The Craziest Base44 Story(12:08) Why Wix Acquired Base44(20:42) Launching Base44's AI Model(30:07) Why You Should Own the Stack(32:33) How We Update AI Models(34:18) What's Still Missing for AI Builders?(39:36) The State of Vibe Coding(47:40) Tech Predictions from Maor & Mark
Our notes from this conversation
* Coding stopped being the bottleneck. What to build is.
Base44 hit $1M in annualized revenue three weeks after launch, then a couple million within three months — solo, with no funding. Maor’s line: “coding is not the issue anymore. The question is what are you going to build.” That’s the thesis for the next few years, not just the next few months.
* Why Base44 and Wix teamed-up
Wix approached Base44 because a “crazy guy” running a one-man show kept generating enough noise that people started telling Mark’s team to buy him. Nir Zohar, Wix’s President, sealed the vibe over steak; the real decision came later, with Maor and Avishai Abrahami (CEO and Founder at Wix) at a whiteboard in the middle of the night, mapping three paths: stay bootstrapped, raise a lot of money, or partner with Wix.
Maor’s real hesitation wasn’t valuation — it was fear of scaling the “magic” too fast and killing the thing that made it work. Avishai’s answer was structural, not financial: lean on Wix for the boring infrastructure, stay lean and move fast everywhere else.
* Owning the front and back end obligates you to own the middleware too.
Base44’s release of Base 1 — its own fine-tuned model for building web applications — wasn’t a marketing move, it was the logical endpoint of a stack Wix and Base44 already owned end to end. Mark’s framing:
If you own the front end and the back end, you eventually have to own how you deliver the service between them, or you’re dependent on someone else’s pricing and roadmap.
* Scale generates the training data that makes owning a model worth it.
Two things moved Base44’s model timeline up by a year or more: open-source models reaching frontier-adjacent quality, and Base44’s own traffic generating millions of usable data points. The lesson generalizes — building your own model only makes sense once you have the volume to fine-tune it well, not before.
* The next differentiator isn’t code quality — it’s taste.
Horizontal coding models are good at many things. Base44’s bet is that a model trained specifically on what makes a good product decision — not just working code — becomes a durable edge as raw code generation becomes commoditized across every vibe coding platform.
* Distribution, not creation, is the unsolved problem for builders.
Mark’s read on the ecosystem:
Plenty of products get built, very few get used, because building well and marketing well are different skills. Base44’s answer is a beta “distribution layer” that uses its own context on each app — B2B or consumer, local or global — to recommend a growth playbook, not just more features.
* Vibe coding is quietly eating enterprise software, not just side projects.
The skepticism about production-readiness a year ago has given way to real adoption: internal tools, dashboards, and niche SaaS products built by domain experts with zero developers on staff. Maor’s example — a former restaurant manager building an invoicing tool for local restaurants — is the shape of a much larger shift in who gets to start a software business.
* Nobody has visibility past about three months, and that’s the honest answer.
Maor won’t pretend to forecast a year out — internal roadmaps only run two or three months because the ground moves that fast. What he will say:
Frontier-model costs keep compressing, open source keeps closing the gap, and the category itself is widening past web apps into games, slide decks, and design — anywhere code can define an output.
* Building in public is now a distribution channel, not a personality quirk.
Maor’s early growth came from documenting the numbers, the features, and the struggles publicly every day, not from a marketing budget. His conclusion for founders generally:
Storytelling and audience-building are becoming as valuable a skill as the product itself, especially for anyone starting without capital.
* A board doesn’t need people like you. It needs people who aren’t.
Mark’s operating principle for Wix’s board — deliberately including outsiders like the CEO of Manchester City for a completely different read on marketing and culture — showed up again when asked who they’d add next: both landed on the same instinct, that the highest-leverage addition is someone who understands modern culture and storytelling, not another operator who thinks like they do.
Links
Follow Ollie on X - https://x.com/ollieforsythFollow Maor on X - https://x.com/MaorShlomoFollow Mark on X - https://x.com/marktluszczVisit Base44 - http://base44.com
Visit Wix - http://wix.com
Previous episodes include
If you enjoyed this episode, help sustain our work by clicking ❤️ and 🔄 at the top of this post.
…..
Brought to you by Harmonic


Get full access to NEW ECONOMIES at www.neweconomies.co/subscribe