The Twenty Minute VCDaniel Khachab: "We Are in the Middle of a Cold War for AI Talent" | E1220
CHAPTERS
- 0:00 – 3:44
Choco’s pivot: why a successful SaaS company went all-in on AI revenue
Daniel explains the sleepless-night realization that LLMs would compress software moats and force companies to disrupt themselves. He outlines the strategic logic behind moving to “AI-first” revenue and upskilling the entire org to build with AI.
- •Fear of tech moats eroding as models learn to code and replicate products faster
- •Two futures: get disrupted vs. disrupt yourself—“always play offense”
- •AI-first as an organizational upskilling race, not just a product feature
- •Employer responsibility: ensure teams learn AI skills to stay relevant
- 3:44 – 5:00
What remains defensible in an AI world: data, relationships, and shifting moats
Harry challenges the idea that moats disappear by pointing to Choco’s distributor relationships and proprietary data. Daniel agrees moats don’t vanish entirely, but argues the technology portion commoditizes quickly, changing what matters most.
- •Distribution relationships and historical data as durable advantages
- •Technology moats shrink as capabilities become accessible via APIs
- •Companies must refocus on non-technical moats and execution speed
- •AI accelerates competitive pressure even in entrenched markets
- 5:00 – 8:34
Jobs and workflows that change first: from UI design to “agent character design”
Daniel predicts many roles will transform as interfaces collapse into prompts and agents. He uses product design and payroll as examples of a shift from feature-heavy UI to conversational orchestration and persona design.
- •UI-heavy workflows (e.g., payroll) shrink into a single prompt + automation
- •Design shifts from screens to agent behavior, tone, and “character”
- •Engineers face new problems like QA for nondeterministic outputs
- •Every function (HR, marketing, sales) must adapt to agent-based work
- 8:34 – 10:44
The agentification of SaaS: why “SaaS is dead” and what replaces it
Daniel lays out the thesis that SaaS becomes “selling an employee,” not selling software features. They discuss what stays (data repositories) and what changes (interfaces, adoption, and how systems like Salesforce are used).
- •“Agentification of SaaS” as the new default software experience
- •Software becomes a flexible workforce: hire 0.1 to 1,000 ‘employees’ instantly
- •Adoption accelerates because users no longer learn complex UIs—just communicate
- •Prompt-led input replaces form-filling; tools win by reducing friction
- 10:44 – 14:07
Enterprise adoption blockers: security, on‑prem models, and the real implementation gap
Harry argues implementation and data readiness are the biggest hurdles. Daniel responds that AI isn’t magic; products must include an interface for training and correction, optimizing for learning rate rather than day-one accuracy.
- •Security/compliance can be addressed with SLMs and on‑prem deployments
- •Implementation pain is real: data cleanliness and post-sale rollout are hard
- •Two interfaces persist: ‘ask the AI’ + ‘train/correct the AI’
- •Success metric shifts to steep learning curves, not perfect initial accuracy
- 14:07 – 16:49
Where value accrues in the AI stack: foundation models vs apps (and model commoditization)
The conversation turns to whether foundation model companies should build applications. Daniel notes rapid price declines and cites Anthropic’s “computer use” as a blurred line between foundational capability and app-layer product.
- •Anthropic ‘computer use’ as a step-change: AI operates mouse/keyboard like a human
- •Foundation vs application layer boundaries are increasingly ambiguous
- •Model prices commoditize rapidly—app-layer players benefit from falling costs
- •Choco prioritizes utility/functionality over price at current maturity
- 16:49 – 23:11
Jobs, layoffs, and labor shortages: does AI destroy work or reallocate it?
They debate job loss versus productivity gains, using Klarna’s customer support cuts as a reference. Daniel frames AI as a response to Western labor shortages, while acknowledging the harshness of AI-driven layoffs and leadership responsibility.
- •Labor shortages in healthcare, childcare, trucking, hospitality as macro reality
- •AI may replace repetitive/undesirable roles and free labor for scarce sectors
- •Daniel describes letting go “hundreds” and the emotional difficulty of AI-driven cuts
- •Leadership framing: mission-first decisions and doing layoffs in one decisive round
- 23:11 – 25:15
The ‘Cold War’ for AI talent: governments competing to relocate teams
Daniel describes governments courting AI founders to move talent across borders, calling it a cold war dynamic. Offers range from visas to extreme salary subsidies for top AI staff, and the strategic trade-offs founders must consider.
- •US, UK, UAE/Saudi actively recruiting AI founders and their teams
- •Relocation incentives: golden visas, operational support, even multi-year salary coverage
- •Talent competition is geopolitical, not just corporate
- •Founders must weigh what’s best for the company and what employees want
- 25:15 – 26:49
Why Berlin for R&D (even with London’s AI gravity): application layer vs foundational layer
Harry questions Berlin as an AI hub compared to London/Cambridge/Oxford. Daniel argues the application layer resets the playing field—teams can upskill quickly—whereas foundational leadership depends on long-established institutions.
- •Berlin advantages: network, livability, affordability, international appeal
- •Foundational AI clusters (DeepMind, major labs) matter more for model research
- •Application-layer AI is new; teams worldwide can learn integration and product design
- •Choco’s R&D concentration: ~15–18 ML engineers in Berlin
- 26:49 – 29:29
Europe’s AI sovereignty problem: chips, energy, and model access arriving last
Daniel argues Europe is structurally disadvantaged: late access to key APIs, minimal chip production, energy constraints, and lack of frontier model builders. He frames these as sovereignty prerequisites for competitive AI ecosystems.
- •Europe receives major AI capabilities later (e.g., voice APIs, Apple Intelligence)
- •Near-total GPU manufacturing concentration in Taiwan; global efforts to de-risk
- •Energy supply as a limiting factor for data centers and compute expansion
- •Sovereignty = chips + energy + foundational models; Europe lacks all three
- 29:29 – 32:43
What Europe can do: attract chip manufacturing, confront energy trade-offs, and close the talent gap
They go factor-by-factor on remedies: fund the right semiconductor winners, solve energy (including nuclear debates), and address the talent and founder-ambition gap for competitive foundation models. The tone shifts to pragmatic, action-oriented proposals.
- •Chips: incentivize credible manufacturers (TSMC/NVIDIA supply chain logic) vs misallocated subsidies
- •Energy: renewables transition is unresolved; nuclear raises intergenerational responsibility issues
- •Foundation models: hardest due to talent density + rare founder commitment
- •Europe needs structural talent strategy and long-term founder ambition
- 32:43 – 36:53
Regulation as excuse vs reality: fragmentation, founder mindset, and ‘stop whining’ pragmatism
Harry and Daniel challenge common European narratives: lack of capital and excessive regulation. Daniel argues capital is global and great founders work around obstacles, citing examples like Tesla’s Berlin factory and US fintech’s regulatory complexity.
- •Capital is accessible via UK/US funds; founders must ‘get on the plane’
- •Regulation can be burdensome (e.g., notarization), but not a showstopper for most SaaS
- •Fragmentation isn’t uniquely European—US has multi-state licensing too
- •Core thesis: great founders adapt stoically and execute despite constraints
- 36:53 – 38:39
Market portfolio strategy: US as dollars, Middle East as ambition, Europe as steady base
Pressed on why not move to the US or UAE, Daniel explains Choco already operates in both and views geography as a diversified portfolio. He contrasts the US’s software scale with Middle East growth and Europe’s more stagnant but durable market.
- •US is largest revenue market and primary investment focus for growth
- •UAE/Saudi offer fast-moving economies with less legacy infrastructure
- •Europe is stable but slower—works well when funded by US success
- •Diversification reduces risk across regulatory and adoption environments
- 38:39 – 43:17
COVID near-death experience: losing 98% GMV overnight and rebuilding by ‘playing offense’
Daniel recounts landing in Berlin to see Choco’s GMV collapse almost entirely. Instead of hibernating, they moved teams to open geographies, shifted GTM from field sales to telesales and PLG, and emerged ahead when lockdowns lifted.
- •Shock moment: 98% GMV drop between Sunday and Monday after travel
- •Personal and organizational strain; crisis required leadership clarity
- •Offense strategy: relocate teams (e.g., Spain team to Miami) to keep selling
- •GTM evolution: field sales → telesales → product-led growth; no mass layoffs
- 43:17 – 46:46
Fundraising, unicorn culture, and the cost of celebrating too early
They discuss the 2023 round at a $1.1B valuation and the trade-offs of taking cash. Daniel’s biggest regret is cultural: the ‘we made it’ mindset that a unicorn label can create, and how it can attract the wrong talent and complacency.
- •Pros of raising: survival through shocks; higher probability of mission success
- •Cons: distraction, reduced capital efficiency, slower execution discipline
- •Unicorn milestone can reduce hunger; celebrating can harm culture
- •Daniel regrets the unicorn party and removed ‘unicorn’ office symbolism
- 46:46 – 55:17
Work ethic, purpose, and founder psychology: intensity over ‘work-life balance’
Daniel rejects conventional work-life balance in favor of being fully present—intense at work and fully present at home. They discuss purpose during crisis, whether they’d do it all again, and the founder trait of embracing repeated setbacks.
- •Mistake in early COVID: giving a week off removed purpose and momentum
- •“100% present” principle: intensity at work, presence at home to prevent burnout
- •Founders must be all-in; if any part is ‘maybe,’ don’t start
- •Learning through falling: parallels from skiing and entrepreneurship resilience
- 55:17 – 1:17:00
Operator lessons: skunkworks failure, Spain as the hardest market, and scaling vs frenzy
Daniel shares practical operational lessons—why special projects teams can undermine core innovation, why Spain is challenging due to fragmentation, and when to accelerate spend. They close with quick-fire beliefs about AI-first companies and smaller teams winning.
- •Skunkworks ‘special projects’ can signal a weak core; innovation must live in the core org
- •Spain is difficult: extreme supplier fragmentation requires different onboarding/GTM
- •When product + GTM economics work, scale fast to build brand and deter copycats
- •Quick-fire: only AI-first companies win; great companies get smaller via automation
