Okay, here are the detailed, narrative-driven show notes for the Hash Rate episode featuring Sergey Volnov of its AI, tailored for Crypto AI investors and researchers.
Show Notes: Hash Rate - its AI (Subnet 32) with Sergey Volnov
This episode explores its AI (Subnet 32), a BitTensor project achieving top-tier accuracy in AI text detection, revealing significant market opportunities compared to traditional VC-backed competitors and the critical role of accessibility in unlocking value.
Introducing its AI: Subnet 32 on BitTensor
- Host Mark Jeffrey introduces Sergey Volnov, founder of its AI, which operates Subnet 32 on the BitTensor network. BitTensor is a decentralized network incentivizing the creation and operation of specialized AI models through its native token, TAO. Subnets are specialized networks within BitTensor focused on specific AI tasks.
- Mark notes his active participation in BitTensor subnets, highlighting his hands-on perspective as an investor.
- Sergey confirms its AI (Subnet 32) is currently ranked around #26 out of ~81 subnets, with a market cap approaching $1 million for its subnet token (Alpha).
The Core Mission: AI Text Detection
- Sergey explains that its AI focuses primarily on detecting AI-generated text. This is increasingly crucial due to the rapid growth and adoption of Large Language Models (LLMs) like ChatGPT. LLMs are AI models trained on vast amounts of text data to understand and generate human-like language.
- The core function of the subnet involves miners distinguishing between human-written and AI-generated text samples provided by validators.
- While acknowledging image/video detection exists (handled by other subnets like BitMind), Sergey emphasizes its AI's strategic focus solely on text detection due to its broader, everyday applicability across billions of users compared to image/video.
Accuracy and Benchmarking: Leading the Pack
- Sergey details its AI's performance journey since registering in March of the previous year. Following a request from Const (a key BitTensor founder, often seen as the "Satoshi" of the project) in September, they benchmarked their model against academic standards.
- On the robust Riot benchmark (launched 2024, with 600k test samples), its AI achieved 95.7% accuracy, ranking #1 globally, even surpassing commercial detectors. Sergey states, "We are currently number one by accuracy among detectors even commercial detectors by this benchmark."
- On other benchmarks (HotpotQA-Free, GhostBuster, GRID), accuracy reached approximately 98%. This high performance is attributed largely to the competitive optimization by the subnet's miners.
Key Use Cases: Beyond Academia
- Mark identifies the obvious educational use case: professors verifying student work.
- Sergey expands on this, highlighting publishers, journalists, and writers who need to ensure content originality.
- Intriguingly, Sergey reveals an "inverse" use case: individuals using AI can leverage its AI's sentence-level probability analysis to identify AI-like phrasing in their text and revise it to appear more human-written, effectively helping users bypass detection systems.
- Mark relates this to Amazon KDP's initial struggle with a deluge of low-quality AI-generated books, underscoring the real-world need for such detection tools. its AI can process very long texts, like entire books, by chunking the analysis internally.
Competitive Landscape within BitTensor
- Mark asks about competitors within BitTensor, mentioning BitMind.
- Sergey clarifies that its AI was the first AI text detection subnet and that BitMind focuses exclusively on image and video detection, meaning they don't directly compete on task. This highlights the specialization occurring within the BitTensor ecosystem.
The Power of Focus: A Strategic Imperative
- Sergey explains the decision to focus solely on text detection was strategic, preventing dilution of effort and maximizing the potential to become the best in that specific niche.
- Mark strongly validates this approach, sharing a powerful anecdote from his time as CTO at Mahalo. Sequoia Capital founder Mike Moritz advised that the only companies with a chance of winning are those hyper-focused on being the best in the world at one thing. Distracted companies, Moritz claimed, inevitably fail. This reinforces the potential strength of its AI's focused strategy.
Subnet Mechanics and History
- Sergey confirms Subnet 32 has 256 slots, nearly full since shortly after launch, with ~240 miners and ~16 validators competing intensely.
- He notes the validation process required several iterations to ensure miner optimizations correlated with real-world accuracy improvements, indicating a maturing process within the subnet.
- The subnet has been operational for about a year, making it one of the older, more established subnets.
Tokenomics and Emission Strategy
- Sergey mentions that historically (during the "pre-Dynamic TAO" phase), the subnet received around 1.8-2% of BitTensor's TAO emissions. Dynamic TAO (DTO) refers to the newer system where subnet token holders (Alpha token holders in this case) vote on TAO emission allocation across subnets, creating a market-driven allocation.
- Currently, under DTO, its AI receives about 0.7% of emissions (approx. 46 TAO/day or ~$12k/day at current prices). Sergey acknowledges the shift towards market dynamics where factors like token buybacks influence perceived value.
- Crucially, Sergey reveals the its AI team has not sold any of their earned Alpha tokens yet. They are funding operations from savings made during the pre-DTO phase and accumulating Alpha to avoid harming the token price. "We currently sponsor all our team... through the savings that we've made during pred stage," Sergey states.
- They recently implemented a minimum Alpha holding requirement for miners to align incentives and encourage long-term commitment, further reducing potential sell pressure.
Market Valuation: BitTensor vs. Traditional AI Ventures
- Mark introduces a critical discussion point: the stark valuation difference between BitTensor subnets and comparable VC-backed AI companies. He uses the example of Scale AI ($14B valuation for human data annotation) versus Ready.ai (a BitTensor subnet doing similar work, ~$6M Alpha token market cap).
- The hypothesis is that BitTensor projects are currently undervalued due to the technical friction and learning curve required to interact with the BitTensor network and Alpha tokens, similar to early Bitcoin/Ethereum before user-friendly interfaces like Coinbase emerged.
Real-World Competition and Market Opportunity
- Asked about non-BitTensor competitors, Sergey identifies key players:
- GPTZero: Raised $10M at a $50M valuation.
- CopyLeaks: Earned ~$5M ARR in 2023.
- Originality.ai: Another significant player (revenue/funding undisclosed).
- Despite these competitors, its AI's superior accuracy provides a strong competitive edge. The existence of funded competitors validates the market demand, which Sergey expects to grow alongside LLM adoption. The current valuations of competitors suggest its AI is operating in a validated multi-million dollar market but is currently valued significantly lower.
Product Access and Future Roadmap
- its AI offers its services via a website (itsai.org) with an application, API access, history, and downloadable certificates.
- Pricing: A premium plan costs ~$50/month or $12/month if paid annually. Enterprise and educational plans are launching soon.
- A major upcoming update includes plagiarism detection, bundling it with AI detection to offer a comprehensive content integrity tool, based on customer feedback primarily from journalists.
- They are also working on more granular highlighting, identifying specific words or phrases most likely flagged as AI-generated to help users refine text.
- The team established a company in Dubai (Its AI Technologies FZCO) in January 2024 to facilitate real-world payments (e.g., credit cards), moving beyond crypto-only transactions.
Meet the Team: Expertise Behind its AI
- Sergey Volnov identifies himself as an AI professional with ~4 years of experience as an ML engineer, previously working in banking and leading an ML department for an LLM-driven real estate tech company in Singapore.
- The its AI team currently consists of about six people: backend, frontend, designer, marketer, Sergey (CEO/ML), and another ML specialist – a lean, focused startup structure.
Future Outlook: Bridging BitTensor to Mainstream Crypto
- Both Mark and Sergey agree that bridging BitTensor subnet tokens (Alpha tokens) to more accessible chains like Solana or Ethereum via wrapped tokens (tokens on one chain representing an asset on another) will be a game-changer.
- This would drastically lower the barrier to entry for average crypto investors familiar with tools like MetaMask or Phantom wallets, potentially unlocking significant liquidity and driving up valuations for fundamentally sound subnets. Sergey notes, "...if you can connect it to like general market like general exchanger it will be much more easier for people to obtain your token..."
Current Bridging Solutions and Associated Risks
- Mark discusses existing solutions attempting to bridge this gap:
- Backprop: Offers a simpler, more user-friendly native interface for TAO and Alpha tokens compared to Taostats.io.
- Taobot: An Ethereum-based platform acting as a centralized custodian. Users deposit ETH, Taobot converts it to TAO/Alpha internally, allowing trading via an Ethereum wallet without directly handling TAO wallets.
- Mark highlights the significant counterparty risk with Taobot: the team is undoxed, and users trust the platform not to abscond with deposited funds (similar risk to early centralized exchanges). He personally tested it with minimal funds but acknowledges its functional ease-of-use proves the concept's appeal.
- Michael White (previous guest, BitTensor fund manager) anticipates Solana-based wrapped token solutions emerging within 1-2 months.
The Binance Analogy: Seizing the Bridging Opportunity
- Mark shares a story about Binance's early growth spurt. By quickly enabling users to receive NEO's GAS token rewards (which other exchanges kept for themselves), Binance attracted a massive influx of users who then stayed.
- The analogy suggests that the platform (whether Solana, Sui, Sonic, etc.) that first successfully and trustworthily implements wrapped BitTensor subnet tokens could experience similar exponential growth by capturing the demand for easier access to these AI assets.
Reflective and Strategic Conclusion
Its AI demonstrates strong fundamentals with market-leading accuracy in a validated niche. The significant valuation gap versus traditional AI highlights a potential opportunity, contingent on improved accessibility via bridging solutions like wrapped tokens—a key catalyst for Crypto AI investors and researchers to monitor closely.
