Built real-time private company valuations that take hours, not months — while staying profitable in an industry where competitors bleed money | Eqvista | Tom Milar | EP 120

The Unlikely AI Pioneer: How heyData Out-Innovates Flashier Silicon Valley Startups | heyData | Miloš Djurdjević | EP 201

The most advanced AI operating model in this episode wasn't built from a strategy deck. It was built from desperation — and that turns out to be the best design constraint available. Miloš Djurdjevic is co-CEO and co-founder of heyData, a Berlin-based compliance platform that scaled to several thousand customers, closed a $16.5M Series A, and kept the team at 60 people — on purpose — while running more than 14 AI agents in marketing, near-fully automated customer support, and more agents than headcount in revenue operations. When heyData was growing fastest before its Series A, the team didn't have a choice. There weren't enough people to handle the customer volume, the compliance questions, or the marketing load. So they started automating — not because it was fashionable, but because the alternative was falling behind. Customer success built a knowledge base from years of accumulated compliance Q&A, then layered AI agents on top until response times dropped from several days to under an hour. Marketing restructured an entire role around managing and iterating on agents rather than producing output directly. Revenue ops rebuilt itself around agents first, humans second. The thread running through all of it: the team was too stretched to be afraid of AI. Every person saw it as capacity relief, not a threat. The secondary conversation in this episode is one Roland flags as increasingly important: whether BI as a function becomes obsolete in an AI-first company. Miloš's answer is direct — at heyData, they depend on it more than ever, not less. As AI takes on more of the analytical work, the job of the BI team shifts from producing reports to ensuring the data underneath those reports is structured, clean, and trustworthy. AI is only as useful as the data that feeds it, and that data hygiene work still requires dedicated human judgment. It's not a disappearing role — it's a fundamentally different one. Roland observes that the heyData story confirms something he sees consistently in his advisory work: durable AI adoption almost never starts with a mandate. It starts with a constraint. The founders who have the most sophisticated AI operations today are, disproportionately, the ones who had no other option a year or two ago. Understanding that dynamic — and what it takes to replicate the results without the underlying pressure — is the open question this episode raises. Key Moments: 00:00 — Why a compliance company became an AI operations lab before AI was fashionable 01:57 — The deliberate case for staying at 60 people after a $16.5M raise — and what every new hire actually costs 05:10 — How heyData built its first AI system in customer success: from a knowledge base of compliance Q&A to sub-hour response times 08:10 — Top-down vs. bottom-up AI adoption — and the cross-functional task force model that made both work together 10:31 — Why heyData's team never feared AI would cost them their jobs — and the specific context that made that true 14:16 — Is BI becoming obsolete in an AI-first company? Miloš's honest answer, and what the role actually becomes 17:21 — Raising a $16.5M Series A in a market that demands AI nativeness — what was harder than expected, and what wasn't 19:31 — 14 AI agents in a 6-person marketing team: how the role of "AI lead" got created and what that person actually does 22:52 — The personal backstory: growing up between Bavaria and Serbia, two doctors for parents, and what persistence looks like when it's inherited If building an AI-augmented operating model — rather than just adopting AI tools — is the challenge in front of your team right now, a Breakthrough Workshop with Midstage is the fastest way to find the right leverage point. Book at breakthrough.midstage.ac #AIOperators #ComplianceTech #AgentFirst #SaaSFounders #BreakthroughAIOperators

The AI Breakthrough Stories Most Founders Are Not Telling | Midstage Accelerator | Doug Miller | EP 200

Most AI adoption failures at growing startups aren't technology problems. They're a founder who is moving fast, surrounded by a team that isn't — and a growing gap between the two that no new tool subscription will close. Roland Siebelink is co-founder of Midstage Accelerator, who has helped scale three unicorns across three countries — each from roughly 10 to 1,000 people in three years — before spending the last decade advising SaaS companies on the hardest years of growth between $1M and $50M. At episode 200, Roland sits down with co-founder Doug to open a new chapter for Scaling Without Breaking: a focused series on what it actually takes to achieve a breakthrough with AI in your operations. The episode starts with a distinction that shapes everything that follows — the difference between using AI and breaking through with AI. Roland's position is clear: AI is a means to an end, and the founders who treat it as the end are the ones who end up more productive and more isolated, running faster while their company catches up. The conversation moves through a telling example — a financial services company that spent years with a six-week manual loan review process, not because the knowledge wasn't there, but because nobody had been able to translate that knowledge into reliable, deterministic code. AI agents changed the equation not by replacing the loan officers but by allowing experimentation that, over time, outperformed the junior hires who had been doing the work. The secondary thread is one that Roland says he sees constantly: the founder who has adopted five or six AI tools, feels dramatically more productive, and cannot understand why their team isn't following. His observation is that the moment a founder prescribes how AI should be used — here's the tool, here's the workflow — they strip away the one thing that makes adoption stick: the team member's own discovery of how AI changes their specific job. The episode closes with Roland describing what Midstage calls the "50% startup" — not a company where AI replaces people, but one where humans and agents each do what the other can't, led by founders who understand the difference. Roland notes that one of the clearest patterns he sees in his advisory work is that the founders who get the most out of AI are rarely the most technical. They're the ones who build the conditions for the team to find the leverage themselves — and who resist the urge to be the chief AI evangelist in their own company. At the $1M–$50M stage, the bottleneck is almost never the tools. It's the change management. Key Moments: 00:00 — Why episode 200 is the right moment to reorient the podcast around AI breakthroughs 01:56 — The difference between using AI and achieving a breakthrough with AI — and why it matters for momentum 03:53 — How an AI coaching agent helped a founder stop dominating team meetings — without a single conversation about it 07:11 — Why an HVAC contractor using an AI phone agent is already outgrowing competitors who can't figure out responsiveness 09:48 — The Venn diagram between AI-native companies and AI operators — and where the real growth is happening 11:25 — The financial services loan company: how AI agents compressed a six-week manual process and what made it stick 16:33 — Why the founder who is "street lengths ahead" on AI tools is sometimes the biggest obstacle to adoption 18:28 — Roland's background: three unicorns, three countries, and what repeated exposure to scaling reveals that first-timers can't see 22:08 — Why external perspective finds the real problem faster — and why founders are often solving the wrong one 26:17 — How to book a Breakthrough Workshop with Midstage and what to expect from it --- Midstage Accelerator is offering founders a Breakthrough Workshop — a facilitated session designed to identify the real constraint in your business and produce a clear path forward. If the gap between your AI adoption and your team's is something you're navigating, this is where to start. Book at breakthrough.midstage.ac #AIOperators #FounderLedGrowth #SaaSFounders #ChangeManagement #BreakthroughAIOperators

From PE Boardroom to Operator — What Investors Get Wrong | HCIM | Ali Evans | EP 122

Most PE investors see a 25-year-old automation business in 2026 and calculate the half-life. Ali Evans saw the one thing AI can't replicate overnight: trust earned over decades of client relationships. The conventional move would have been to exit before disruption. He acquired the company instead and stepped into the CEO chair. Ali Evans spent years at Francisco Partners and Riverside writing checks and sitting on boards before leaving the investor seat to acquire HCIM, a healthcare automation company founded in 2000. He's now CEO of the firm he owns at the fund level, navigating the rare dual role of PE investor and operating executive. The first challenge Ali walked into wasn't technical — it was linguistic. Founders and investors use the same vocabulary but optimize for completely different outcomes. Founders want to solve more customer problems. Investors want to maximize return per unit of risk. That gap destroys trust in almost every PE-backed deal, and Ali found himself on both sides of it simultaneously. The moment he became CEO, he had to stop speaking in investor math and start speaking in the language his team actually cared about: impact, autonomy, and customer outcomes. The conversation also covers why Ali doubled down on healthcare automation in the age of AI, why he thinks relationships and trust are harder to replicate than technical expertise, and the story behind Metamora — his firm's name, which comes from the hardest week of his life at a small-town football camp that taught him what it means to bond a team through crucible moments. Roland sees this founder-investor language barrier break down trust in almost every SaaS deal he advises on at the $5M–$20M stage. The misalignment isn't usually strategic — it's definitional. Both sides think they're talking about growth, but the founder means "how many more customers can we help" and the investor means "what's our IRR on this capital deployed." Ali's experience living in both seats at once confirms what Roland keeps telling both founders and their backers: you're not disagreeing on the plan. You're disagreeing on whose calculus gets to define success. Key Moments: 01:58 — Why AI can mimic expertise but can't replace 25 years of earned trust — and why that's the bet Ali made 05:33 — The Spanish vs Portuguese problem: how founders and investors use the same words but mean completely different things 08:01 — The exact moment Ali realized his team didn't care about risk-adjusted returns — and why he had to unlearn investor-speak to lead 12:45 — Why most founder-investor conflicts aren't strategic disagreements — they're fights over whose definition of success matters 16:30 — How Ali transitioned from writing checks to running the company — and the due diligence he did on himself before becoming CEO 21:10 — What agentic AI workflows are actually doing in healthcare automation — beyond the marketing buzz 25:22 — The Metamora origin story: how a brutal high school football camp became the name of Ali's firm — and what it taught him about bonding teams through hard things HCIM is offering listeners a free workflow automation readiness assessment report. If you're running a healthcare payer operation and wondering where RPA could save real operational time, reach out at https://hcim.com/contact/ [https://hcim.com/contact/] If bridging the gap between founder-speak and investor-speak is something you're navigating at the $5M–$20M stage, Midstage works directly with SaaS CEOs to translate strategy into execution without losing either side. mdstg.ac/drag-erase [http://mdstg.ac/drag-erase] #PrivateEquity #HealthcareAutomation #FounderInvestorDynamics #OperatorCEO #ScalingWithoutBreaking

Why 20 People Can Beat Billion-Dollar AI Companies | Deep Infra | Nikola Borisov | EP 121

Open source AI models are now just 3-5% behind the best closed source models on benchmarks — about six months of lag time, not five years. If you're building an AI infrastructure company on the assumption that OpenAI or Anthropic will maintain a permanent lead, your moat is disappearing faster than your revenue projections assume. Most founders at the $3M–$20M stage are still over-indexed on model selection and under-indexed on inference economics. They're obsessed with training costs and model access, but the real cost explosion is coming from running models at scale. A model that trains for a year but only runs for a month is a terrible investment — and yet that's how most AI budgets are still structured. Nikola Borisov spent a decade building backend infrastructure for a chat app with 200 million monthly users before launching Deep Infra. He's CEO and co-founder of Deep Infra, an AI inference platform that owns its own GPU clusters and serves as one of the largest token suppliers on OpenRouter. The episode centers on two bets Nikola made that most infrastructure founders won't: first, that open source models would catch up to closed source models faster than anyone expected, and second, that inference — not training — would dominate AI budgets within five years. Those bets are both paying off. The gap has narrowed to 3-5%, and as Deep Infra lowers costs, customers aren't just consuming more tokens — they're jumping to better, bigger models. The conversation also surfaces a less obvious pattern: the economics of AI inference mirror the economics of CDNs more than they mirror cloud compute. Walmart and Target don't care if their images are served from the same CDN — it's just an efficient way to deliver content. Deep Infra runs the same model for multiple companies in parallel on the same GPUs, and neither company cares. It's neutral infrastructure that scales horizontally without requiring every company to build their own. Roland sees this pattern constantly in his advisory work with SaaS companies scaling from $1M to $50M: founders are modeling their AI spend around closed source API access and per-token pricing, but they're not accounting for what happens when open source closes the gap and inference costs drop 20x. The companies that move early to open source inference infrastructure will have a cost structure their competitors can't match in 18 months — and cost structure at scale is the actual competitive wedge, not model access. Key Moments: 3:01 — Why the gap between closed source and open source models has narrowed to 3-5% — and what that percentage actually measures 5:00 — The five-year-old explanation of inference: training is school, running the model is work 6:41 — Why Anthropic's compute conflict (training vs. serving customers) reveals the real economic wedge 10:39 — The CDN analogy: why Walmart and Target don't care if their requests run on the same infrastructure 16:12 — How lowering costs changes customer behavior — they jump to bigger models, not just more tokens 18:51 — Why Nikola believes inference will dominate company budgets in 5-10 years 20:29 — What a math Olympiad medalist and programming competitor learned about certainty that still drives how he builds 22:31 — Nikola's advice to younger founders: focus on what's most important today, not what's interesting --- If navigating AI infrastructure economics — balancing model access, inference costs, and long-term vendor lock-in — is something you're working through right now, the Midstage Accelerator helps SaaS founders at the $1M–$50M stage model these decisions with real unit economics and stage-specific benchmarks. mdstg.ac/drag-erase #AIInfrastructure #OpenSourceAI #InferenceEconomics #SaaSScaling #ScalingWithoutBreaking

Built real-time private company valuations that take hours, not months — while staying profitable in an industry where competitors bleed money | Eqvista | Tom Milar | EP 120

Every cap table company in Silicon Valley is burning venture capital chasing growth. Tom Milar built one that makes money instead — and he did it without ever raising a pre-seed round, despite managing $300 billion in client assets for companies including Perplexity AI. The question this episode refuses to let go of is whether the VC-fueled growth playbook has become so normalized that founders have forgotten there's another way to build. It turns out the founders closest to the problem — the ones who see thousands of cap tables and valuations up close — might have the clearest view of what's actually driving company value.   Tom is the founder and CEO of Eqvista who built and sold the largest incorporation provider in Hong Kong, acquired the largest registered agent in Nevada, and then built a profitable valuation and cap table platform serving 23,000+ companies — all without a single successful fundraising round.   The core of this episode is Tom's "European Hong Kong pragmatism" — a phrase that sounds like a geographic contradiction but is actually one of the most coherent operating philosophies in the episode. Where Silicon Valley rewards the pitch, Tom rewards the product. His argument is direct: the first version of everything — the code, the pricing, the privacy policy, the initial client agreements — has to come from the founder. Not because delegation is wrong, but because no hired CTO or product manager can build what they don't deeply understand. Eqvista's website was, by Tom's own description, hideous for six years. He left it that way deliberately. When he eventually changed the design, conversions didn't move. Neil Patel, who served as board president at one of Tom's acquired companies, had told him the same thing: you can kill a business by touching what's already working. The insight Tom keeps returning to is deceptively simple — build a hierarchy of problems by asking which ones are closest to revenue, and work down from there. Everything else is noise.   The secondary thread is what Tom observes watching founders up close through Eqvista's platform, where he sees not just cap tables but the underlying behaviors of companies raising seed rounds, Series A, and beyond. His diagnosis of why founders move too slowly isn't about effort — it's about sequencing. The founders who fail are the ones who raise $400,000 before doing the heavy lifting themselves, who hire before they understand the product, who feed the beast with venture money before the product is strong enough to stand on its own. The real-time valuation product Tom demos in the episode crystallizes what's at stake: the two factors that drive private company valuations fastest are revenue and how well a founder can sell equity. Everything else — the design, the brand, the org chart — is downstream of those two numbers.   What Roland observes repeatedly at the $1M–$50M stage is that the founders who stay profitable longest tend to have built an intuition about revenue proximity that their VC-backed peers often lack — not because they're more talented, but because they've never had the option of substituting capital for clarity. The discipline Tom describes, building a hierarchy of problems anchored to what makes money, is something most founders only develop after they've run out of runway once. The companies that arrive at Midstage having bootstrapped or stayed lean tend to have sharper product instincts and messier org charts; the ones that raised heavily tend to have the reverse. Neither is inherently better, but knowing which problem you have is the first step to fixing it.   Key Moments 00:39 — Why Tom couldn't raise a pre-seed round despite having multi-billion dollar companies as clients — and why he's not sure it mattered 02:32 — "European Hong Kong pragmatism": what it actually means to build profitable companies in a culture that rewards hype over product 04:45 — Why the first code, pricing, and client agreement must come from the founder — and what gets lost when founders skip that step 08:25 — Two types of founders Tom sees up close: the ones who can raise, and the ones who can build — and why the ideal is rarer than anyone admits 10:02 — The real-time valuation demo: what a living stock price for a private company looks like, and why fund admins processing 500 companies traditionally need six people for six months to do what Eqvista does automatically 13:33 — The one filter Tom uses to cut through every meeting, every email, every decision: what makes money? 14:36 — Why Eqvista's website stayed hideous for six years — and what happened to conversions when they finally fixed it 16:29 — The founder sequencing trap: why raising $400K before doing the heavy lifting yourself is one of the fastest ways to fail 18:01 — The two factors that drive private company valuations the fastest — and what every founder building toward a raise needs to understand first   ---   Eqvista is offering Scaling Without Breaking listeners $100 off their plan. If you're a founder managing equity, planning a raise, or wanting a real-time valuation for your company, this is the most direct way in. Use referral code ROLAND_EQVISTA at eqvista.com.   If you're navigating how to structure equity, understand your company's true value, or think clearly about what a raise would actually do to your cap table, Midstage Institute works directly with SaaS and software founders at the $1M–$50M stage to help you make those decisions with clarity before they become expensive mistakes. mdstg.ac/drag-erase   #EmployeeStockOptionPlan #PrivateEquity #VentureCapital #SeedRound #ScalingWithoutBreaking