Ventura Labs

In this episode, we are joined by Sergey Volnov, CEO of It's AI and a Bittensor subnet owner, who shares the progression of his career from competing in international computer science Olympiads to developing enterprise AI text detection systems. He discusses the specific demand for detection tools caused by the increased use of large language models, the qualitative shortcomings of early detection software, and how a decentralized validation system maintains the accuracy of detection models against new language models. Sergey also details the mechanics of his subnet, how miners are incentivized to perform precise text segmentation, the strategy of targeting B2B markets in the Middle East to compete with GPTZero, the business rationale for keeping core algorithms closed-source, and his operational focus on text detection over expanding into other data types.

In this episode, you’ll learn

• Sergey’s technical background, from winning international computer science Olympiads to managing machine learning departments in Singapore focused on natural language processing
• The architecture of his subnet, including how validators utilize verified pre-AI text and outputs from current language models to force miners to update their detection algorithms
• How It's AI requires miners to provide granular text segmentation instead of binary classification, and how this data integrates into learning management systems like Moodle
• The methodology behind industry evaluations like the MGTD and RAID benchmarks, and the data indicating how It's AI achieves the accuracy to compete with platforms like GPTZero and Turnitin
• The strategic decision to maintain closed-source intellectual property to preserve the business model, and the execution of direct B2B enterprise sales in the MENA region
• Sergey's operational rule on the necessity of maintaining absolute focus on a single, specialized problem rather than dividing technical resources across image or video detection

In this episode, Ventura Labs interviews Aaron Basch, CEO and Co-Founder of SPARKET and the team behind Subnet 57, sharing his journey from pricing Settlers of Catan games with friends to building GLI-certified betting infrastructure for massive brands like Station Casinos and Penn Entertainment. He covers how SPARKET differs from platforms like Polymarket by focusing purely on the data and infrastructure layer, the glaring flaws in traditional sports data validation, and how his 2020 patent for crowdsourced data validation perfectly aligned with Gittensor's decentralized architecture. Aaron also dives into the mechanics of Subnet 57, how miners are rewarded for beating traditional sportsbook consensus, the challenges of migrating a Web2 company to a Web3 ecosystem, navigating complex US gambling regulations, and practical advice for traditional tech companies looking to build on blockchain.

In this episode, you’ll learn

• Aaron’s origin story: from pricing video game bets during COVID to developing a parimutuel engine used to facilitate wagers for local golf tournaments and the National Lacrosse League
• How SPARKET acts as a pure infrastructure and data layer compared to exchanges like Polymarket, and why the current industry standard of relying on a single official to validate game outcomes is a massive vulnerability
• The mechanics behind his newly granted 2020 patent for crowdsourced sports data validation, how it assigns "credit scores" to accurate data providers, and why discovering Gittensor was the perfect catalyst to build it
• The inner workings of Subnet 57: how miners provide probability distributions for European soccer and are scored based on final outcome accuracy and how early they provide the signal
• Navigating the transition from a traditional Web2 software company to an open Web3 ecosystem, including why they intentionally slowed down their development cycles and borrowed auditor designs from the Schuetz subnet
• The complex landscape of state-by-state US gambling regulations, how SPARKET is structuring its Alpha token to provide value without crossing regulatory boundaries, and Aaron's core advice for existing companies looking to integrate blockchain

In this episode, Ventura Labs interviews Jared Lawrence of Beyond Finance, a trading and investment community focused on Bittensor, sharing his journey from hotel management and business optimization to full-time trader. He discusses early binary options losses sparking risk management passion, building Bittensor conviction via Siam's community, Beyond Finance's holistic philosophy blending mindset, health, and finance, TAO price insights amid manipulation, the QE bomb and liquidity cycles, low-liquidity subnet trading, astrology's market role, CBDC and stablecoin concerns, and advice for subnet traders in this early ecosystem.

In this episode, you’ll learn

• Jared’s backstory: entrepreneurial roots, managing a hotel at 21, running a cleaning business to fuel investments, and diving into Bittensor’s incentive mechanisms
• Trading origins: starting with binary options (losing 5K in two weeks), evolving to risk-managed oil derivatives during split shifts, and shifting focus to TAO accumulation over portfolio value
• Building Bittensor conviction: joining Siam’s community in September 2023 at $100-150, reallocating his entire portfolio to subnets and TAO stacking
• Beyond Finance essence: beyond trading signals, a holistic mindset, nutrition, exercise, and emotional control approach, making up 80% of trading success
• TAO price suppression: manipulation, macro cycles, and being "too goddamn early," plus how the QE bomb (10-20T printing) will pump crypto assets
• Newcomer pitfalls: ignoring risk management (1-2% per trade), and Jared’s teachings on position sizing, stop losses, and alerts for long-term gains
• Bittensor trading strategies: low-liquidity subnets with slippage awareness, perpetual futures on mega-tile decks, and embracing new subnet volatility with risk control
• Jared’s Bittensor explanation: "decentralized Silicon Valley" for retail, fractional AI startup stakes via subnets, with incentives for constant improvement
• Taoflow as indicator: on-chain buying momentum, initial large-cap favoritism, and net positive impact despite challenges
• Macro views: 3-5 year asset tokenization timelines, stablecoins as dollar-saving Trojan horse, CBDC dystopian control risks, and QR code crypto payments future
• Astrology in trading: edges like new moon dumps and full moon pumps, viewing 2024 as transformative opportunity amid volatility
• Subnet trader advice: personalized systems by pot size and risk tolerance, chart-focused entries/exits, TA with fundamentals, and treating it as a trader’s market pre-liquidity surge

In this episode, Ventura Labs interviews Doug Sillard of Taostats, Bittensor’s most patient DevRel, community manager, and educator, sharing his accidental journey from aspiring professor to full-time crypto. He covers his AT&T DevRel roots, how Mogg recruited him for Corcel docs at 32 subnets, the evolution of his educational videos, the 167-endpoint Taostats API, chain buys mechanics, Dynamic TAO one-year review, MEV bots and slippage, TaoFlow challenges for new subnets, root claim variance, unsolved accessibility problems, the 128 subnet cap, and advice for subnets explaining complex tech.

In this episode, you’ll learn

• Doug’s backstory: aspiring professor to accidental DevRel at AT&T, Mogg’s outreach for Corcel docs (subnets 18 & 19), and learning Bittensor by teaching it through still-live early videos
• Early Bittensor: zero knowledge of miners/validators/subnets at 32 subnets, pre-Dynamic TAO, and compartmentalizing the firehose of concepts
• Taostats docs & API: 167 endpoints, why running your own 3+ TB archive node is rarely worth it, and building custom tables/wallet history
• Team growth: front-end, designers, backend, and Subnet 19’s Block Machine becoming the new indexing backend
• Chain buys explained: EMA of TAO flows, capping emissions, recycling excess TAO, and why Doug sees it as a smart post-halving solution
• Dynamic TAO one-year review: compromise design that worked, emission tweaks, excess TAO handling, and what’s still evolving
• MEV bots: how they front-run stakes, slippage limits that mostly stop them, and why many users still get wrecked by old extrinsics
• TaoFlow for new subnets: what actually counts (only staking/unstaking), positive flow requirements post-halving, and the needed TAO backlog strategy
• Root claim variance: shift from every 72 minutes to ~36-hour average (sometimes 10 days), minimum alpha thresholds, and why the change helps dust claims
• Biggest unsolved problem: making Bittensor accessible to average traders amid too many complex options and 128 subnets
• Subnet cap future: will 128 ever come off, current deregistrations, and startup-like 90% failure rate in the ecosystem
• Advice for subnets: explain your tech so clearly that listeners can retell it to others, distill AI complexity into understandable stories

In this episode, Ventura Labs interviews Gordon Frayne, a prolific crypto content creator with 12K+ YouTube subs, sharing his shift from digital innovation studies and e-commerce to full-time crypto in 2024. He covers 2017 Bitcoin entry, ICO vs. subnet parallels, Tao mindset shift, subnet trifecta, staking/DeFi correlations, Dynamic Tao impact, buy-and-hold strategy, Tao-pilling tactics, bull case amid dips, incentive layer explanation, awareness solutions, Yuma consensus, and ecosystem contribution tips.

In this episode, you’ll learn

• Gordon's backstory: 2017 ICO boom entry, e-commerce scaling, daily 2024 YouTube commitment for Tao awareness growth
• ERC-20 vs. subnets: Evolved fixed-supply model in Tao for competitive accrual and decentralized AI innovation
• Tao trading mindset: Bitcoin-inspired accumulation via yields and outperformers during 2026 sub-$150 dips
• Subnet trifecta: 30-150% APY staking, Tao outperformance, long-term dollar growth to thousands
• Subnet staking vs. DeFi: Uniswap/Orca parallels, impermanent loss, set-and-forget simplicity, research focus
• Dynamic Tao review: Net positive for democratized AI equity, despite updates and extractive risks
• Buy-and-hold investing: DCA roots, framework assessing models, incentives, teams, burns
• Bridging ecosystems: Indirect Tao-pilling from Ethereum/Solana, bull case on undervalued fundamentals
• Explaining Bittensor: Incentive layer for proof of intelligence, TauStats for aligned mining
• Awareness fixes: Content, events, incentives as natural selection, talent network with 30-50K daily rewards
• Yuma consensus: Useful work efficiency over Bitcoin's waste, subnet-specific incentives like Shoots compute or Carta liquidity
• Contribution advice: Three-prong staking/mining/trading for two-way value in incentive layer

In this episode, Ventura Labs interviews Seby Rubino, founder of Resi Labs AI (Bittensor Subnet 46), which uses AI models for real estate appraisals, inspections, and tokenized transactions to revolutionize systems like MLS and mortgages. He covers the subnet's pivot from data to predictions, MLS tokenization, 99% accuracy via inspections, "mortgages you don't pay back," Resi Finance liquidity, REIT differences, asset scaling, full-stack roadmap, and big-vision subnet advice.

In this episode, you’ll learn

• Seby's real estate data roots and subnet history: Bittensor excitement, quick pivot post-Const/BT Labs input for miner-led AI innovation
• Subnet design: Miners submit Hugging Face models for daily sold-property predictions, enabling verifiable lender/investor feedback
• Pivot rationale: Implicit data scraping via models, roadmap acceleration, miner talent as IP for productization
• Real estate "BS" critique: MLS outdated, thesis on brokerages as tokenized platforms automating trust, slashing fees to 0.1-0.5%
• Transaction essentials: Appraisal/inspection/title, agentic layers for niche data, 95-99% accuracy surpassing Zillow/humans
• Traditional flaws: Post-agreement inspections/renegotiations, liquidation-focused appraisals; AI enables instant P2P efficiency
• Tokenization vision: Shared ownership for non-repayable mortgages, appreciation/rental yields, Resi Finance pooling for leverage
• Strategies: 50% LTV borrowing on Airbnb tokens for 10-20% yields, empowering users beyond centralized REITs with liquidity
• Scaling: Fractional models for cars/planes, Resi expansion to Inspect.Network for broad value inspections
• Endgame: Models/agents for intelligence, oracles/contracts for infrastructure, liquidity for finance; lender revenue now
• Timeline: Appraisals/inspections by year-end, liquidity 2027; roadblocks like concept proof; subnet tips: Question norms, think huge, backward from visions with market-fit steps

In this episode, Ventura Labs interviews Gavin Zaentz and Pranav Ramesh, co-founders of LeadPoet (Bittensor Subnet 71), which builds a decentralized system for high-quality leads using miners and validators to surpass tools like Apollo or ZoomInfo. They discuss their NASDAQ meeting, Bittensor discovery, rapid growth to 1.1M leads, double-validation, intent signals, ecosystem partnerships, sales automation roadmap, and subnet launch tips.

In this episode, you’ll learn

• Gavin and Pranav's backstory: Meeting at NASDAQ, Gavin's Helium and early Bittensor involvement, and their conviction in Bittensor after events like Endgame
• LeadPoet's origin as a fix for lead gen pains in past startups, using Bittensor miners for fresh, high-intent leads scaling to 50K/day
• Explaining Bittensor via "contributors" and incentives, akin to ImageNet's open-source revolution in image recognition
• Subnet mechanics: Miners rewarded for validated leads (emails, LinkedIn, roles); validators ensure quality
• Qualification processes: Intent from social/PR/funding, e.g., cybersecurity buy/sell signals for private equity
• Data mapping/verification: Linking platforms via websites/URLs to confirm authenticity and block fakes
• Double validation: SMTP checks at entry/delivery to prevent bounces, excluding catch-alls for better deliverability
• Nvidia Inception perks: Compute, discounts, networks for AI startups, cutting costs without equity loss
• Partnerships with subnets like Bitssec, Metanova, Chutes: Testing data, priority for agents/outreach
• Prior venture frustrations: Manual sifting, high bounces inspiring AI intent detection on Bittensor
• Roadmap: Automate sales from leads to bookings, letting founders focus on selling
• Usage: Burn TAO for credits, ~3¢/lead initially, open-source for white-labeling
• DSV incubation: Ideation to subnet slot, capital, networks in a pioneering partnership
• Subnet advice: Strong validation, tackle barriers with decentralized compute, build on personal pains
• Community shoutout: Expo events, sales help offers, highlighting Bittensor's collaborative vibe despite glitches

In this episode, Ventura Labs interviews Marc, founder of Numinous, the team behind Bittensor Subnet 6, which fosters a competitive ecosystem for AI forecasting agents using open-source code, genetic algorithms, and winner-takes-all incentives to drive superhuman prediction accuracy. Marc discusses his background in pure maths at Cambridge, the pivot to an open-source agent competition model, the preference for unified forecasting models over specialized ones, the LLM approach for handling complex predictions like those on Polymarket, the end goal of building an "Oracle of Truth" with explainability, performance benchmarking, available tools like Chutes and Desearch, innovative agent designs, the winner-take-all structure, and advice for standing out in Bittensor's competitive landscape.

In this episode, you’ll learn

• Marc's journey from Paris to Cambridge for pure maths, his shift from a PhD to crypto and Bittensor, and how a London hacker house accelerated his work by fostering community and collaboration
• The major pivot in Subnet 6 from traditional forecasting to an open-source agent competition, enabling transparency, unified models, and faster innovation through shared code and an efficient frontier of best practices
• Why Numinous favors one big, versatile forecasting model over routing to smaller specialized ones, allowing agents to handle diverse domains like crypto, macroeconomics, and geopolitics effectively
• The advantages of an LLM-based approach for Polymarket predictions over pure time-series models, especially for incorporating real-time news, satellite data, and complex events that cause market jumps
• The dual end goal of creating a superhuman "Oracle of Truth" for accurate probabilities while achieving explainability through reverse-engineering inputs, sources, and reasoning processes
• How performance is judged by outperforming Polymarket odds, with upcoming benchmarks, simulated P&L, and autonomous trading agents to execute on forecasts
• Insights into benchmarking challenges, including liquidity variations, execution layers for trading, and plans for a Polymarket fund to validate real-world beating of markets
• What makes Subnet 6's design feasible for prediction markets, emphasizing general forecasters, constrained environments, and confidence in miners achieving superhuman levels without news access initially
• Details on tools like Chutes for LLMs and Desearch for X APIs/web search, with roadmap additions like Gemini, Perplexity, and curated high-value data sources for specific markets (e.g., live sports APIs)
• Surprising agent designs emerging, including genetic algorithms for continual optimization, architecture exploration, and miners re-optimizing for event types to build on baselines
• The rationale for a winner-take-all incentive structure to drive efficiency, with potential future tweaks to reward ensembles or marginal improvements for broader diversity
• Marc's advice for differentiating in Bittensor's crowded prediction market sector: Focus on your unique vision, ignore competitors, and let the market validate through sincere problem-solving and steps toward a forecasting world model
• A glimpse into handling disruptions like background noise during the interview, underscoring the raw, community-driven spirit of Bittensor builders

In this episode, Ventura Labs interviews Keith Singery and Garrett Oetken (Carrot) from Tensora/TAO.com, known as the "Carrot and the Stick" from their new podcast. As chief evangelist and head of protocol respectively, they discuss the merger forming Tensora, Keith's validator journey, TAO.com's role in onboarding new users, evolving validator and subnet owner responsibilities, subnet design shifts, governance and trust mechanisms, TAO Flow's impact, handling exploits, the ideal balance of hardware vs. skill in mining, fostering subnet collaborations, and a live demo of the TAO.com wallet app, plus advice on building trust in the ecosystem.

In this episode, you’ll learn

• The origins of the "Carrot and the Stick" podcast and how it bridges technical insights with ecosystem evangelism for better Bittensor understanding
• Details on Tensora's structure post-merger, including co-founders' roles and TAO.com as a professional, user-friendly gateway out of the crypto echo chamber
• Keith's transition from independent validator to joining Tensora, leveraging his stake for broader ecosystem value and professional tool development
• Why TAO.com stands out as the easiest interface for new users to access subnets and escape the Bittensor bubble
• The pivot from validators building products to subnet owners taking the lead, due to focused incentives and the challenges of scaling validator operations
• Contrasts between old and new subnet designs, including access models, miner incentives, and the value of sticking to original Bittensor principles for alpha tokens
• Validators' refined future role in governance, trust verification, and decision-making on alpha vs. TAO, plus alliances influencing staking choices
• How Bittensor's ethos has shifted from traditional mining to efficient models like miner burns, treasuries, and capacitors for better capital use
• The evolution of subnet focus from pure research pre-Dynamic TAO to revenue-driven commoditization post-launch, while all remain "research subnets" at core
• Strategies for unlocking higher-stack intelligence commodities beyond basic infrastructure like inference, compute, and storage
• The ideal miner balance emphasizing skill over pure hardware to produce novel intelligence, with infrastructure subnets as foundations for skill-based ones
• Examples of subnet stacking (e.g., Templar, Gradients, Shoots) and how an "orchestra" of collaborations relieves individual product pressure
• Addressing internal tensions amid price dips, with external demand and leader integrations driving cross-subnet partnerships
• Positive views on Dynamic TAO and TAO Flow as refinements prioritizing inflows over price for fairer emissions, despite initial shakeups
• Handling self-staking "exploits" through market corrections and viewing every exploit as an opportunity to strengthen the protocol
• The role of social enforcement in governance, aiming for trust, transparency, and community voice to harden Bittensor for scale in 2026
• A demo walkthrough of the TAO.com wallet app, highlighting easy staking, subnet discovery, buying with Face ID, and info-rich subnet pages
• Security features like Secure Enclave storage and non-custodial keys, plus the app's overlooked value in centralizing subnet updates for better visibility
• Advice for identifying trustworthy actors: Seek personal interactions, attend events like Exploit, and focus on fundamentals like incentives over surface changes

In this episode, Ventura Labs interviews Maciej Kula of Latent Holdings and founder of learnbittensor.org. This is Maciej's second appearance on the podcast, where he discusses his role in maintaining Bittensor docs, evolving LearnBittensor into a guided academy, common misconceptions like chain buys, detailed mechanics of halvings and alpha injections, the recent TaoFlow changes, MEV solutions, proxy wallets for security, root claim updates, and advice for deepening Bittensor knowledge through fundamentals.

In this episode, you’ll learn

• Maciej’s transition to Latent Holdings and the vision for LearnBittensor as an accessible, guided learning platform starting from crypto basics
• Key differences between developer-focused docs (with SDK examples) and LearnBittensor’s beginner-friendly breakdowns to reduce redundancy and improve onboarding
• Common overlooked Bittensor features, such as chain buys existing since February, and how they create buy pressure through alpha recycling
• In-depth explanations of alpha downscaling, pool injections, and how chain buys adjust prices without inflating tokens excessively
• The estimated first TAO halving around December 15th, why it’s not bearish for participant rewards, and its effects on liquidity growth rather than incentives
• Separate halving schedules for TAO (affecting pools) vs. alpha (affecting rewards), including caps on injections and fixes for compression issues
• Maciej’s positive stance on TaoFlow, how it solves price-based emission problems by focusing on inflows, and EMA lookback periods for fair calculations
• Remaining challenges like MEV front-running and the upcoming encrypted transaction solution to protect against sandwiches and slippage
• Benefits of proxy wallets for secure staking/unstaking without exposing cold wallets, plus types like owner proxies, small transfers, and root claims
• Upcoming root claim defaults shifting to "delegate" for validators to decide keep/swap on alpha dividends, and potential outflows impacting emissions
• How root proportion decreases with less TAO on root or more alpha issuance, influencing subnet stakes
• Reflections on Bittensor’s rapid evolution as the "fastest-moving blockchain," inspired by community involvement and constant updates
• Advice for reaching expert-level understanding: Focus on fundamentals like incentive mechanisms and game theory rather than surface-level changes
• Maciej’s emphasis on questioning core purposes (e.g., incentivizing TAO inflows) to adapt to shifts like from price-based to inflow-based emissions