Ridges AI (SN62)

Overview

Ridges AI is a project on Bittensor that develops AI agents for software engineering. Its subnet, SN62, runs competitions in which developers submit coding agents and validators test how well they solve programming problems. The aim is to improve agents that can handle a complete task, rather than help with just one step of writing code. Ridges also offers Ridgeline, a product that puts a coding agent to work on GitHub issues. (docs.ridges.ai)

SN62 has its own alpha token within Bittensor. The token helps coordinate rewards for agent developers, validators, and people who stake on the subnet. This connects the project’s two sides: a competition that produces better agents and a tool that lets software teams use them. (docs.ridges.ai)

History & Team

From AgenTao to Ridges

The subnet launched in the fourth quarter of 2024 under the name AgenTao. It later became Ridges AI. During 2025, the project expanded its coding benchmarks and changed its policy for publishing successful agent code. In the first quarter of 2026, Latent acquired Ridges, and Ridgeline launched in beta as a tool for assigning software tasks to agents. Later updates added more focused competitions and new ways to pay for Ridgeline jobs. (docs.ridges.ai)

Shakeel Hussein identifies himself as the person who built Ridges. The project’s public code repository also shows ongoing work by multiple contributors. DSV Fund publicly disclosed an investment in SN62 made through an over-the-counter token allocation and market purchases. Those purchases concern the subnet token; they do not, by themselves, describe an ownership stake in the company. (shakeel.com)

Technology & How It Works

A contest for coding agents

A coding agent is software that uses AI models and tools to work through a task. In Ridges, a developer—called a miner—builds an agent and submits its Python code to a competition. Different competitions test different skills. Ridges lists niches for linting, database query engineering, and test generation. An agent enters one competition and is judged against other entries in that same niche. (docs.ridges.ai)

The agent moves through screening stages before it reaches validators. At each stage, it works on problems inside an isolated environment. It submits a code change, or patch, which is applied and checked against tests the miner cannot see. For validator-stage problems, the platform awards credit only when all assigned validators agree that the problem was solved. This makes the result depend on working code rather than an AI model’s opinion of the answer. (docs.ridges.ai)

Ridges uses hidden and varied problems to test whether agents can apply a skill beyond one familiar example. It also screens submissions for behavior designed around known answers or test details. When an agent leaves its leading position, its code is published so others can study and build on it. The platform coordinates submissions and evaluations, while Bittensor records the reward weights used to distribute subnet emissions. (docs.ridges.ai)

Tokenomics & Utility

What the SN62 token does

Bittensor has a network token called TAO. Each subnet also has its own alpha token; SN62’s token belongs specifically to Ridges. Under Bittensor’s protocol, each subnet alpha has a maximum supply of 21 million units. Alpha enters the system through the network’s emission process, with part directed toward subnet participants. It should not be confused with TAO or with the alpha of another subnet. (bittensor.com)

Ridges uses emissions to reward approved agents. When several competitions run, each receives an assigned share. Within a competition, qualifying agents divide rewards according to scores that reflect how much they improved on the leader when approved. An agent may qualify by solving more problems or by matching the leader’s performance at a meaningfully lower inference cost. Its reward score then fades over time, giving developers a reason to keep improving their work. (docs.ridges.ai)

Alpha has uses beyond rewards. Miners register a Bittensor hotkey to submit agents, and an upload consumes a fee paid in SN62 alpha that is burned. People can also stake on the subnet. Ridgeline grants job credits to eligible users who lock their staked SN62 alpha and connect the relevant Bittensor wallet. Simply holding the token or staking without the required lock does not earn those credits. (docs.ridges.ai)

Ecosystem & Use Cases

From benchmark to GitHub task

The most direct use of Ridges’ work is Ridgeline. A user connects a GitHub repository and provides an issue that describes the desired change. The agent sets up the project, reads the task, edits code, runs checks, reviews its changes, and opens a pull request. A person can then review that pull request before merging it. This gives the project a practical outlet for agents developed through its competitions. (docs.ridges.ai)

The competition niches point to more specific tasks. A linting agent works on code that must pass analysis rules. A database-focused agent handles code that queries a live database. A test-generation agent writes tests meant to catch bugs. These are useful skills for a broader coding agent, even though success in a competition and performance on a user’s repository are different measures. (docs.ridges.ai)

Ridgeline jobs can be funded through a subscription, credits earned from locked subnet stake, or an x402 payment for an individual job. The x402 route uses USDC on the Base network. That payment method serves the product, while SN62 alpha serves the Bittensor subnet and its staking-based credit system. (docs.ridges.ai)

Advantages & Challenges

Ridges brings independent developers into a shared contest with defined tasks and tests. Its scoring gives participants a clear target, and its reward rules recognize both better results and lower costs. Publishing former leading agents also lets later entrants learn from earlier approaches. For users, Ridgeline turns that development work into a familiar GitHub workflow: describe an issue, receive a proposed change, and review it. (docs.ridges.ai)

The design also sets demanding goals. A miner must build an agent that works across hidden problems, fits within the evaluation process, and improves enough to qualify for emissions. Ridges has to keep its tests useful as coding agents improve. On the product side, Ridgeline’s results depend partly on how clearly a task is written and whether the repository can be set up and tested from a clean checkout. The project continues to work on the path from a GitHub issue to a dependable pull request. (docs.ridges.ai)

Where to Buy & Wallets

SN62 is available on Kraken. It can also be acquired through Bittensor’s subnet staking system by exchanging TAO for subnet 62 alpha. These are two ways to access the same subnet token, though exchange account features and direct on-chain staking work differently. (blog.kraken.com)

A Bittensor wallet can be managed with btcli, the network’s command-line tool. Talisman is a browser wallet that supports Bittensor, and Ridges lists it as an example for connecting a wallet to Ridgeline. The connection uses a signed message to link a Bittensor address to a Ridgeline account. That link is part of earning product credits from locked SN62 stake. (docs.ridges.ai)

Regulatory & Compliance

SN62 activity sits within the rules that apply to digital assets and the services used to trade them. In the United States, the IRS treats digital assets as property for federal tax purposes. Sales, exchanges, mining rewards, and staking rewards can have reporting consequences. In the European Union, MiCA sets authorization requirements for businesses providing covered crypto-asset services. These frameworks address different parts of the ecosystem: a token’s use on Bittensor, an exchange’s services, and a user’s tax reporting are separate matters. (irs.gov)

Ridges’ published product and token documentation describes staking, rewards, burns, and payment options, but does not give a formal Shariah or halal ruling for SN62. Its halal status therefore remains undetermined in those materials. A religious assessment would need to consider the specific activity involved, including whether someone is using the product, developing agents, or receiving staking rewards. (docs.ridges.ai)

Future Outlook

Ridges’ stated work in progress centers on improving Ridgeline’s GitHub workflow, making its agent more reliable, and adding focused competitions for engineering skills the product needs. This direction links the subnet’s incentives more closely to tasks developers want completed. The project has already moved from a single broad contest toward multiple niches that can run alongside one another. (docs.ridges.ai)

The lasting test for Ridges is whether improvements measured in its competitions carry over to useful work in real repositories. Its structure provides a way to pursue that goal: miners try new approaches, validators measure results, and Ridgeline gives users a way to put coding agents to work. (docs.ridges.ai)

Summary

Ridges AI combines a Bittensor coding-agent competition with a GitHub-based software tool. SN62 rewards participation in the subnet and connects locked stake to Ridgeline job credits. Its place in the crypto ecosystem comes from using token incentives to develop AI agents, then applying those agents to software tasks people can review and use. (docs.ridges.ai)