Max Hodak: Average Is Not Good Enough

• Max Hodak is the CEO of Science, a deep-tech startup focused on brain-computer interfaces (BCIs) and medical devices.
• Science's primary product is a retinal prosthesis: a chip implanted under the retina to restore vision to patients blind from rod/cone loss.
• The device utilizes a solar cell grid (hexagonal pattern) where an external camera/laser system projects images in infrared to excite the retina directly.
• The product completed major clinical trials last year, was featured on the cover of Time magazine, and has enabled a patient to read a 300-page novel.
• The company operates in a deep-tech space requiring significant physical infrastructure, contrasting with pure software startups.

Infrastructure and Purchasing Efficiency

• Universal Startup Challenge: Deep tech founders often underestimate the need to buy thousands of physical components, unlike pure software founders.
• Credit Card Limitations: While founders can use credit cards initially, scaling requires formal procurement systems for 17+ employees to avoid bottlenecks and burn control issues.
• The "Cheap Power Supply" Fallacy: Spending time negotiating a 50% discount on a $3,000 power supply is inefficient if it delays a high-paid engineer by a week; the labor cost far outweighs the hardware savings.
• Procurement Bottlenecks: Traditional vendor account setups (insurance, certification, POs) can cause a two-week delay for simple orders; active management and dedicated purchasing staff are required to maintain speed.
• Speed through Internal Software: Science built Helix, an internal software platform where almost every action (including purchasing) is a button click, correlating lab steps with costs to prevent delays.
• Cost Attribution: In deep tech, experiments often feel "free" because materials are shared in bulk; accurate pricing requires tracking specific costs (e.g., a single wafer iteration costs $40,000).
• Runway Reality: For a 20-person deep-tech company, headcount and facility costs (e.g., $20k sq ft at $4/sq ft) consume the majority of the budget, often depleting a Series A faster than anticipated.

Hiring and Performance Review Systems

• Nucleation Strategy: The best hires come from the "scene" or extended network that birthed the company, as they already share the necessary language and context.
• Defined Process: Successful startups require a rigid, religiously followed hiring process to prevent recruiting from becoming a bottleneck.
• Company-Wide Voting (Helix): Applications trigger a vote from 7-8 employees matched to the candidate's background to average judgment and avoid small-group bottlenecks.
• Screening Metrics: The phone screen evaluates three traits: Judgment, Horsepower, and Agency (the ability to make a complex situation better).
• AI-Resistant Homework: Science prioritizes tasks with high ceilings and low saturation (e.g., GPU optimization) that are naturally scorable and difficult for AI to fake.
• Conversion Targets: To avoid wasting time, interviews should have a minimum 25% conversion rate to offers; lower rates indicate a broken funnel.
• Eigenreviews System: Instead of annual 360 reviews, the company uses a graph-based voting system every 4-6 weeks asking, "Would you vote to hire this person again today?"
• PageRank Analogy: Eigenreviews use eigenvector centrality (weighted by the quality of the voter) to generate continuous, unbiased performance signals, detecting anomalies via Markov chain Monte Carlo dropout.
• Speed as Success: The core thesis is that rate of iteration separates winners from losers; a company learning one new fact weekly will outcompete one learning monthly regardless of the competitor's other advantages.

Founder Judgment and Leadership

• Non-Delegable Judgment: Founders cannot delegate final decision-making; they must make decisions that make sense to them personally, even when alone in their conviction.
• Action Generates Information: When stuck, founders must inject "action" to create entropy and new information, rather than waiting for theoretical solutions.
• Oral Tradition: Skills in organization and judgment are best learned by working for high-performing companies before founding one's own; there are no generic algorithms for success.
• PhD vs. Industry: If a field is dominated by basic research, a PhD is reasonable; once the field matures, industry (with greater resources and speed) overtakes academia in execution.
• Evidence of Ability: Exceptional ability is best proven through "legible competitive games" (e.g., Formula SAE, chess) rather than abstract credentials.
• AI in Hiring: Science evaluates thinking and problem decomposition over specific coding skills (LeetCode), as software rewards horsepower and rapid feedback loops.
• Neuralink Lessons: Working with a leader of "empirically excellent judgment" helps founders refine their own decision filters for high-stakes, long-term bets.

BCI, AI, and Industry Outlook

• BCI Goal: The endgame of BCI is not just super-intelligent machines, but conscious machines that can participate in the loop with humans.
• Consciousness Research: Proving consciousness is a practical engineering problem requiring a mapping between neural substrate activity and phenomenal content, likely requiring human trials.
• Interdisciplinary Teams: Smaller, integrated teams (e.g., protein engineering solving electronics constraints) outperform large, siloed academic centers by compressing the problem into fewer minds.
• AI Impact Areas: AI has transformed Science in coding (reducing developer burden), regulatory compliance (automating standards tracking and evidence generation), and general R&D.
• Quality Systems: AI helps navigate complex regulations by automating the creation of evidence tables for standards (e.g., battery safety, shipping labels) that are otherwise burdensome for humans.
• Build vs. Buy: Founders should consider building custom internal software (like Helix) because off-the-shelf ERPs rarely fit specific needs; "vibe coding" now makes this feasible for seed-stage companies.
• Longevity Narrative: Hodak views BCI primarily as a longevity/healthcare story (fixing the brain directly) rather than an AI story, citing dramatic effect sizes (e.g., restoring vision) unattainable in pharma.
• Implant Challenges: Key engineering hurdles include power/thermal constraints, wireless power delivery, and packaging (protecting the implant from the immune system/bio-fouling).
• Cultural Balance: Successful companies (like SpaceX and Science) need a mix of mission-driven "lunatics" (30%) and serious engineers (70%) to drive radical innovation while maintaining execution.

Fundraising and Strategy

• Experiment-Based Raising: Founders should calculate the exact cost to run the next value inflection point experiment and raise twice that amount to account for waste and discovery.
• Revenue Focus: Even deep-tech companies must prioritize revenue to shift valuation from "probability of dying" to "long-term roadmap."
• No General Principles: There are no shortcuts or pithy instructions for startups; success relies on refined, domain-specific judgment and the willingness to think for oneself.
• Power over Money: The most effective founders are motivated by the "power" to change the world (e.g., healing) rather than economic gain.
• Biotech Reality: Biotech is capital-intensive and difficult, requiring a decade-long commitment, but offers a unique ability to share power (health) compared to military or economic power.