Why your ‘Great Idea’ Is quietly killing your startup
with Mike Vitez, Saturnia Design
Why your ‘Great Idea’ Is quietly killing your startup
Show Notes
Mike Vitez is a machine learning-trained computer scientist and co-founder of Saturnia Design, a product development studio that works with founders at the messy interval between scrappy MVP and legitimate scale. His background spans ML algorithm development, multiple startups (including an NFT metaverse platform he exited during the 2021 boom), and years of watching the same expensive mistake repeat itself: founders fall in love with their idea, skip user validation, build the wrong thing at full cost, and then act surprised when the market doesn't show up.
His core conviction is blunt: execution is 99% of startup outcomes. The idea is 1%. And the most common execution failure is not technical - it is epistemic. Founders don't ask their users.
The Biggest UX Mistake: Not Asking
When Mike is asked what he sees most often go wrong with user experience, the answer is immediate: founders don't ask their users. And when pressed further: 99% of the time, the problem is not what the founder thinks it is. Not even close.
This is not a condemnation - Mike is consistently emphatic that founders should be founders: hustle, vision, dream. That is what they are for. But the low-level professional work of extracting honest signal from users - watching them struggle, asking the right questions, sitting with uncomfortable answers - is a different skill set. And it is a skill set that is easier with distance. Sub Scan, the Polkadot-backed blockchain scanner, is a representative example. The system was technically robust. The team knew it was hard to use. Saturnia's role was to find exactly which pain points were most blocking users - not to redesign everything, but to prioritize ruthlessly so the engineering effort landed on what actually mattered.
The Third-Party Truth Audit: Why Founders Can't Do This Alone
Mike's signature engagement - what he calls a UX consulting project or UX audit - is built around a structural truth about human psychology: founders cannot get real feedback themselves. Two distortions reliably corrupt the data.
First, customers are gentle with founders. They do not want to hurt the feelings of the person who built something and is clearly passionate about it. They soften the criticism, emphasize the positives, leave out the part that would actually help. Second, founders are defensive with customers. When you hear something broken, you feel exposed. You want to explain, contextualize, defend. You hear what you can manage, not what is actually being said.
A neutral third party removes both distortions. The result: an honest, prioritized action list. These are the blocking bottlenecks costing you revenue. These are secondary issues. These can wait. Not features you feel proud of - bottlenecks you can now actually see and do something about.
Don't Build from Scratch: The Engine Principle
One of Mike's strongest operational positions: in 2025, building from scratch is immature. Not ambitious - immature. The right move is to find the existing engine closest to what you need, build around it, customize it, connect it to other systems. The work is integration, not invention.
This applies to no-code and low-code tools (Lovable, Squarespace, GoHighLevel), to open-source frameworks, and to existing platforms that can be extended rather than replaced. The engineering leverage is in knowing which engine to choose and how to connect the components - not in writing everything from zero.
The implication for team structure: you need engineers, not coders. Coders put bricks; engineers design the house. Bricks can be automated. Architecture cannot. Mike's studio deliberately hires fewer coders and uses AI in their place, reserving human engineering talent for decisions that require genuine design judgment.
The Three Converging Capabilities
Mike identifies three capabilities whose combination is reshaping how products get built: existing engines (platforms, open source, frameworks), NLP (natural language AI that can be prompted rather than programmed), and automation (trigger-based API integration - workflow automation that has existed for 10–15 years and is now being commercialized at scale). Connect these three and you have a product development stack that requires far less custom code and far more product thinking.
His prediction: within two to four years - likely closer to two - this three-part combination will reach enterprise-level commercialization. That is where disruption of legacy enterprise software actually happens: not when a startup out-codes the incumbent, but when a smaller team can compose equivalent capabilities from existing parts at a fraction of the cost.
When Not to Build: Mike's Four Red Flags
Mike's four conditions for walking away from a founder engagement are worth writing down. First: the founder is too ambitious to be realistic - they repeat unachievable targets and stop listening when told so. Second: they want a two-year build timeline before going to market. In 2025, two years is a lifetime; a product with a two-year launch horizon will be irrelevant before it ships. Third: they won't put skin in the game - ideas and hard work are the bare minimum, not a value proposition. Fourth: they have not thought about jurisdictional distinctions - the legal and regulatory differences across markets that have forced large players with millions spent to fail market entries they assumed were straightforward.
- The Third-Party Truth Audit - Have a neutral outsider interview users, observe them using the product, and deliver a prioritized action list. Customers soften feedback for founders; founders get defensive with customers. A third party removes both distortions and surfaces the honest bottlenecks blocking revenue. See Frameworks.
- Engine + NLP + Automation - Modern product development converges three capabilities: an existing engine (platform, open-source framework, no-code tool), natural language AI, and trigger-based automation. Build around the right engine and connect components rather than coding from scratch. In 2025, building from zero is immature. See Frameworks.
- Pre-MVP Concept Validation - Before writing code, test the concept: a landing page, a single function, or direct customer questions. Confirm that the problem is real and that users want to interact with it the way you imagine before committing engineering resources. Most founders skip this and build the wrong thing at full cost. See Frameworks.
- Lovable - AI-powered web app builder that lets founders go from prompt to deployed product without custom code. Mike cites it as an example of the engine category: an existing platform to build around rather than rebuild from scratch.
- Figma - Design tool used by Saturnia for wireframing and high-fidelity UI design. Mike sees Figma-to-code pipelines as one of the key near-term product development unlocks: design once, export to working code with minimal engineering overhead.
- Saturnia Design - Mike's product development studio. Works with founders at the MVP-to-scale stage: UX research, user interviews, product strategy, wireframing, design, and development. Acts as an external product team when founders don't have the in-house capacity to staff all disciplines.
- Execution Premium - The principle that execution accounts for 99% of startup outcomes and the initial idea for 1%. Ideas are the starting point; validation, iteration, and delivery quality determine whether anything becomes a business. See Glossary.
- UX Audit - A structured research engagement in which a third party interviews users, observes product usage, identifies pain points, and delivers a prioritized action list that separates blocking bottlenecks from secondary issues. Distinct from an internal review: the third-party framing removes the dual distortions of customer courtesy and founder defensiveness. See Glossary.
- Engineer vs. Coder - Engineers are system architects who decide what to build and how components fit together; coders implement those specifications. Engineers are irreplaceable; coders can increasingly be replaced by AI. A critical distinction for founders deciding how to staff a technical team - hire for architecture, not for bricklaying. See Glossary.
- Pre-MVP Concept Validation - Testing a product concept before building an MVP: a landing page, a waitlist, a single-function prototype, or direct customer interviews. The goal is to confirm that the problem is real and that users want to interact with it in the way you imagine - before committing engineering resources. See Glossary.
- Jurisdictional Risk (Product) - The legal, regulatory, and cultural constraints that differ across markets (US, EU, Asia, Australia) and that fundamentally shape what a product can do in each. Founders frequently fail to account for these before entering new markets, resulting in expensive late-stage pivots. Even well-funded players have failed market entries by underestimating local distinctions. See Glossary.
- Market Timing - The alignment between a product's launch and market conditions that make adoption natural. Mike's NFT metaverse startup succeeded in part because timing was right; the same product would have failed 18 months earlier or later. A component of startup success that is not fully controllable but is retrospectively knowable - and prospectively worth analyzing before committing to a category. See Glossary.
- DAO (Decentralized Autonomous Organization) - A governance structure in which decisions are made collectively by members using transparent, often blockchain-based voting mechanisms. Mike advocates for DAOs as the infrastructure for hyper-local charitable giving: communities vote on their most pressing problems, and funding is algorithmically allocated based on local scoring rather than by centralized NGO decision-making. His Donaton concept applies this model to grassroots philanthropy. See Glossary.
Why do founders fail to get honest feedback from their own users?
Two structural distortions. First, customers are gentle with founders - they don't want to hurt the feelings of someone who has clearly poured themselves into something. They soften the critique, emphasize the positives, and omit the part that would actually help. Second, founders are defensive with customers - when you hear something broken, you feel exposed, you want to explain. You hear what you can manage, not what's actually being said. A neutral third party removes both. The feedback is harder to receive and more useful to act on.
What is a UX audit, and what does a founder get out of it?
A UX audit is a structured research engagement: a third party interviews representative users, observes them navigating the product, extracts patterns, and delivers a prioritized action list. The output has two layers. A narrative that explains what was found and the reasoning behind the priorities - the CTO reads this to understand the architecture of the problem. And a task list, ranked by business impact, distinguishing the blocking bottlenecks costing you revenue from the secondary issues that can wait. Sub Scan, the Polkadot-backed blockchain scanner, is a real example: technically solid, hard to use, and Saturnia's job was to identify exactly which pain points mattered most before the redesign.
Why is building from scratch immature in 2025?
Because there are already excellent engines available for virtually every product category - open-source platforms, no-code tools, established frameworks - and the engineering leverage is in choosing the right one and connecting it to what you need, not in recreating it from zero. The founders who succeed fastest find the closest existing engine, customize it, and ship. The ones who insist on writing everything themselves spend 12 months catching up to where they could have started. Mike's formula: find the engine, add NLP, connect automation. That combination is commercially available now and sufficient to build most products that founders actually want to build.
What is the difference between an engineer and a coder, and why does it matter for hiring?
Engineers design the system - they decide what to build, how components fit, what problems the architecture needs to solve. Coders implement the design - they put the bricks in the house. The distinction matters because bricks can be automated. Coders are increasingly being replaced by AI that executes specifications competently and cheaply. Architects cannot. Founders who staff their technical team with coders instead of engineers end up with a lot of fast execution and no coherent architecture. Mike's studio deliberately hires fewer coders and uses AI in their place, keeping human engineering capacity for genuine design decisions.
When should a founder not build a product?
Mike gives four conditions. One: the founder is too unrealistic to update - they repeat unachievable targets after being told they're unachievable, and they stop listening. Two: they want a two-year build timeline. Two years is a product lifetime in 2025; the market will move past it before launch. Three: they won't put skin in the game. Ideas and hard work are the bare minimum, not a value proposition. Four: they haven't thought about jurisdictional distinctions - the legal and regulatory differences across markets that have forced large, well-funded players to fail on market entries they assumed were straightforward. Any one of these is a reason to have a very direct conversation before starting.
What are the hot product categories Mike would build in today?
Three. Mental health and wellbeing - rising demand, under-served by current products, strong consumer motivation. Healthcare - same thesis as Diagnostic MD: fragmented, high-cost, ripe for consumer-centric disruption. And Web3 infrastructure problems that look solved but aren't - specifically, the on-ramp/off-ramp flow with KYC in a single, coherent user journey. He's conspicuously bearish on AI as a product category for anyone who isn't a large player: the success rate of AI-native startups born in the last two or three years is already declining, and the moats belong to the companies with the compute and data. Consultancy around AI adoption he exempts from this - that market is large and underserved.