Conductor

What do they actually do

Conductor is a Mac app that lets developers run multiple coding agents in parallel, each in its own isolated copy of a real Git repo on the user’s machine. It handles repo cloning and git worktrees so you don’t have to juggle branches or directories, and shows what each agent is doing at a glance (site, docs).

Today it supports Claude Code and Codex agents, uses your existing Claude login or API key, and includes a diff viewer plus a workflow to review changes, create PRs, run checks, and merge from one place. There’s basic issue integration (e.g., Linear) to spin up workspaces from tickets (site, YC profile, docs: workflow).

The product runs locally on macOS and is currently free; the team says they plan to charge later for team collaboration features (docs FAQ).

Who are their target customer(s)

• Individual Mac software engineers and indie hackers experimenting with AI assistants: Managing multiple agent sessions and branches is tedious; they want a quick way to spin up parallel agents against real repos without manual cloning or session juggling so they can ship code faster.
• Small feature teams trying to parallelize tickets: Work stalls on coordination and review; they need visibility into each agent’s progress and a clean path from agent output to PRs and merges.
• Engineering managers/tech leads focused on quality and auditability: Agent-produced changes can be unpredictable; they need an interface that makes progress visible, keeps work isolated from main, and turns changes into reviewable diffs/PRs.
• Developers maintaining brownfield codebases: Context is fragile; they need agents to work in isolated copies of the real repo and respect project rules so generated changes don’t break things.
• Teams already using Claude Code or Codex who want to scale agents: Running many agents is tedious; they need orchestration that runs multiple agents in parallel and handles git/workspace setup automatically.

How would they acquire their first 10, 50, and 100 customers

• First 10: Work with trusted early users from the founders’ and YC networks via white‑glove installs. Pair live sessions on the user’s repo with rapid fixes, then turn each successful run into a short case note and quote.
• First 50: Publish a handful of copy‑paste “agent recipes” and short videos, list on developer discovery channels, run weekly group demos, and add a self‑serve trial with a sample repo. Do targeted outreach to indie-hacker forums and GitHub maintainers and convert with 30–60 minute assisted trials plus a simple referral incentive.
• First 100: Run structured two‑week pilots with small teams led by a customer‑success hire, ship templates/integrations for common workflows, and publish 3 short case studies. Layer in developer marketing (how‑to guides, editor plugins, meetup sponsorships) and a lightweight sales playbook to convert pilots into paid seats.

What is the rough total addressable market

Top-down context:

There are roughly 19.6M to 47.2M developers globally, depending on methodology. About 31.8% primarily develop on macOS, and roughly half of professional developers report using AI tools daily (JetBrains 2024: 19.6M, Statista/Evans: 28.7M, SlashData 2025: 47.2M, SO 2024 macOS 31.8%, SO 2025 AI daily 51%).

Bottom-up calculation:

Mac devs = global devs × 31.8%: low 6.2M (19.6M×31.8%), mid 9.1M (28.7M×31.8%), high 15.0M (47.2M×31.8%). Daily‑AI Mac devs (near‑term adopters) = Mac devs × 51%: ~3.2M to ~7.6M addressable users (SO 2024, SO 2025, JetBrains, Statista/Evans, SlashData). Dollar context: AI code tools are a multi‑billion segment projected to ~USD 26B by 2030 (Grand View Research).

Assumptions:

• macOS share (primary dev OS) approximates the reachable install base for a Mac‑only app.
• Daily AI users are the near‑term buyers most likely to need agent orchestration; usage correlates with willingness to pay.
• macOS share and AI‑tool usage are treated independently across the developer population.

Who are some of their notable competitors

• Goose: Open‑source, local‑first coding agent (desktop + CLI) that executes and tests tasks in isolated workspaces. Overlaps on “run agents locally with real repo/CLI access.”
• Claude Code (Agentrooms): Anthropic’s multi‑agent workspaces and desktop client. Teams can coordinate multiple agents directly, reducing the need for a separate Mac orchestration layer.
• Cline: IDE‑ and CLI‑centric coding agent with VS Code/JetBrains/terminal integrations for developers who prefer agent help inside their editor rather than a separate app.
• Kiro: Developer‑focused autonomous agent (desktop + CLI) that structures repeatable agent workflows on local repos; overlaps where teams want planned, agent‑driven feature work.
• Junie (JetBrains): JetBrains’ AI coding agent with IDE‑first flows, multi‑file edits, and VCS assistance; competes for teams that want AI embedded in JetBrains tools instead of a separate Mac app.