Reason
Reason Machines builds Reason, an autonomous software engineer/coding agent for engineering teams. It connects to codebases, reproduces bugs, writes and tests code, and produces review-ready pull requests with recordings and logs.
At a Glance
- Software engineering teams
- Startups and enterprise software organizations
- Developers needing cloud or local coding-agent execution
- Teams using GitHub-based development workflows
AI Tools by Reason
(1)
Reason Network
Peer to Peer Coding Agent Messaging
Discussions
No discussions yet
Be the first to start a discussion about Reason
Latest News
Introducing Reason Agent 1.1; Reason reports higher Coding Agent Index performance than Codex at lower cost.
Reason launches on Product Hunt as a cloud-or-local coding agent available from Mac, terminal, web, and Slack.
Meet Reason for macOS, with local repository/tool access and the ability to continue sessions elsewhere.
Reason Machines listed by Y Combinator as an active Spring 2026 developer-tools/enterprise-software company building Reason Agent.
Products & Services
Autonomous software engineer/coding agent that takes engineering tasks, accesses a repository and computer, writes code, runs tests, and opens evidence-backed pull requests.
Reason's agent harness release built on Pi, emphasizing parallel tool execution, a minimal general-purpose tool set, cache eligibility, and lower cost per solved task.
Desktop app that connects a Reason session to a user's Mac, including a selected checkout, installed tools, local files, and development ports; sessions can also be continued on the web or iOS.
Open-source engine behind reason network: a team's coding agents talking to each other, with peer-to-peer and MCP-oriented capabilities. The reason-machines repository is published under Apache-2.0 and MIT licenses.
Market Position
Reason positions itself as a cost-efficient autonomous coding agent and harness, competing with Codex, Pi, Devin, Claude Code, and other coding-agent/AI IDE products. Its stated differentiation is an agent that can work cloud or locally, maintain a separate persistent computer per session, run tools in parallel, and produce tested pull requests with proof; its October 2026 benchmark post reported a 65.1 Coding Agent Index score versus Codex's 62.5 and 22% lower cost per solved task on GPT-6 Astra.
Leadership
Founders
Adi Singh
Founder/CEO; studied EECS at UC Berkeley and Cybernetics at NTNU, worked in applied ML at Scale AI, and previously built a Rust-based IDE from scratch.
Sven Myhre
Founder/CTO; studied EECS at UC Berkeley, worked as an ML engineer on production systems/scaled ML infrastructure, chaired Norway's largest student AI organization, and won first place in the Norwegian AI Championship in 2025.
Executive Team
Adi Singh
Founder and CEO
EECS at UC Berkeley and Cybernetics at NTNU; applied ML at Scale AI; previously built a Rust-based IDE.
Sven Myhre
Founder and CTO
EECS at UC Berkeley; ML engineer on production systems and scaling ML infrastructure; former chair of Norway's largest student AI organization.
Founding Story
The founders started with the vision of an IDE that runs the development loop end to end rather than waiting for instructions. Their earlier Ara concept responded to the shifting bottleneck in software development—from writing code to reviewing and planning—by proactively proposing and executing the next step; Reason focuses that direction on getting from a task to a tested pull request while allowing the agent to keep working independently of a user's laptop.
Business Model
Revenue Model
Reason Cloud offers free and paid subscription plans. Local use can run on a user's own computer and users can provide their own API keys or use credentials managed through Reason Cloud; cloud sessions are subject to subscription-plan limits.
Pricing Tiers
Free local use is advertised; the terms also state that Reason Cloud offers a free plan.
Paid Reason Cloud plans renew by billing period; plan-specific prices were not shown in the fetched public sources.
Target Markets
- Software engineering teams
- Startups and enterprise software organizations
- Developers needing cloud or local coding-agent execution
- Teams using GitHub-based development workflows
- Delegating bug reproduction and fixes
- Implementing features and writing regression tests
- Running codebases, browsers, terminals, and local development services
- Generating review-ready pull requests with evidence
- Parallelizing engineering work across multiple coding-agent sessions
- Autonomous software engineering for startup and enterprise development teams