Jared Lawrence

AdTAO: Decentralized Ad Optimization on Bittensor x.com/i/broadcasts/1…

The Nerds hosted @jill_kenney from $TAO @sundaebar_ai for an AMA. Here is what we found. The Bittensor ecosystem has no shortage of subnets talking about enterprise agents. A lot of it is still just narrative. After hosting @jill_kenney from SN121 in The Nerds, we came away thinking @sundaebar_ai may actually be building something real. The idea is ambitious and obvious at the same time: build a generalist AI agent for business. Something that works across teams, handles workflows end to end, and becomes part of how a company operates. Big vision, yes. But the more important question is what exists today. That part was better than we expected. Today, @sundaebar_ai has a live marketplace with agents, prompts, workflows, and as of tomorrow, skills. They also have a custom design studio for businesses that want agents built now. The marketplace is doing more than sitting there looking pretty. It is helping them test demand, gather feedback, and push those learnings back into SN121 challenges. That loop between what businesses want and what the subnet trains for is one of the most interesting parts of the model. There is also some early traction. @jill_kenney mentioned multiple enterprise builds, a $30K contract announced, $10K booked in March, and another $20K in inbound. All inbound so far, minimal outbound push yet. These are founder-reported numbers, not independently verified, but if they hold up they are meaningful for something still this early. What we respected most is that they pivoted. The earlier modular approach sounded like it could have turned into chaos fast, what @jill_kenney called death by a thousand agents. Instead, they simplified the model. One main product, then a growing library of reusable skills and components that can be assembled for different business needs. That is really what SN121 is training for now. Not one-off agents, but reusable enterprise skills that can be composed into real workflows. Retrieval, reasoning, structured execution, research, memory, outreach, file handling, tool use. Improve those core pieces and you improve the broader agent over time. After eight challenges and more than 1,300 agents tested, that seems to be the lesson they took from the data. Challenge 9 is where things will get more interesting. Instead of miners submitting full agents, they will submit skills directly. That is a meaningful shift. It ties the subnet more directly to products that can actually be sold through the marketplace. The basic logic is pretty clean: challenges create useful skills, those skills go into the marketplace, marketplace activity creates revenue, and that feeds back into the broader product. The evaluation side may end up being one of the most valuable parts. @jill_kenney described multiple graders, LLM-as-judge, and datasets with metadata and ground truth built in. The point is not just to say whether an agent worked. It is to judge performance with nuance. An agent that hits a word count but misses key details should not score the same as one that slightly misses the count but gets the substance right. That kind of evaluation matters. As more agents get built, the ability to test properly becomes a real business in itself. The broader moat, in their view, is Bittensor. The decentralized competition model is supposed to solve what most off-chain competitors cannot: general enough to adapt across enterprise workflows, but continuously improving through open market incentives rather than a single internal team. That thesis still needs more stress-testing, but the architecture at least points in the right direction. It is still early. Still needs more proof. But after this AMA, @sundaebar_ai looks a lot more serious than we expected. They are not just talking. They are building the marketplace, the skill library, the eval layer, and the commercial loop all at once. If they execute, SN121 could end up being one of the more interesting agent subnets on the network.

I am pleased to announce that Stillcore Capital @stillcorecap has invested in RESI @resilabsai (Bittensor Subnet 46).

damn @Sebyverse is talking real estate on @twistartups when @jason is speaking... $TAO Bulls are listening 👂 Figure Heloc is trading at a $15B FDV... @resilabsai is trading at $10M Could this be the Zillow killer??

Bittensor & Subnets: A New Era x.com/i/broadcasts/1…

.@handshake_58 skills now integrated into our marketplace. subnet x subnet = compounding capability 🤝 🍨 Link in comments.

Bittensor $TAO's @0x_Markets SN35 is building one of the most asymmetric opportunities. Hyperliquid fees are brutal, $450 USDC in fees on a single $BTC close. Same mission permissionless access to the world's biggest markets. Decentralized perpetuals on FX, Crypto, and Commodities. Seven live markets day one: BTC/USD, ETH/USD 100x leverage XAU/USD Gold, XAG/USD Silver 200x leverage EUR/USD, GBP/USD, USD/JPY 500x leverage Five hundred times leverage on FX pairs. On-chain. Permissionless. This is the first subnet offering real foreign exchange and commodity trading with this kind of leverage directly on decentralized infrastructure. And the timing is perfect. Hyperliquid fees increasing. 0xMarkets is aiming straight at that pain point with decentralized liquidity that actually works. The testnet competition running right now tells you everything about their confidence: 22,000 Alpha in the prize pool (roughly $30K+ USDC equivalent) Top 10 traders by PnL take home real Alpha 500K USD0 starting stack per wallet 1 month duration They're literally paying traders to stress-test the platform before mainnet. The subnet metrics: TVL: $737K and growing Weekly Rewards: $23,436 (97.06 $TAO) Average lock duration: 316 days stakers are locking for the long haul Daily Emissions: 2,232 alpha ($3,348) The APY structure is designed to reward conviction: 30 days: 15.2% APY 90 days: 45.7% APY 180 days: 91.4% APY 365 days: 185.2% APY Lock longer, earn more. Protocol-level alignment for long-term believers. This is the subnet playbook executed correctly: Build the product. Run a real testnet with real incentives. Ship mainnet. Let the market decide. No vaporware. Pure trading performance with on-chain settlement. The vision: "If it has a price, you'll be able to trade it on 0xMarkets." This is a multi-trillion dollar addressable market. Alpha price is $1.50. Mainnet is weeks away. Locks. Real revenue is about to start flowing. The market hasn't priced this yet. @0x_Markets $TAO Of course, not advice. DYOR.

Synth Q1 Alpha Call x.com/i/broadcasts/1…