← Back to research
·8 min read·company

Conductor

Conductor is a YC-backed Mac desktop app for running parallel Claude Code, Codex, and Cursor agents in isolated git worktrees, with a $22M Series A.

Key takeaways

  • Mac-native app for running multiple coding agents in parallel on isolated worktrees
  • Supports Claude Code, Codex, and Cursor; uses your existing subscriptions (no additional model costs)
  • Local-first: all code stays on your machine, agents work in git worktrees
  • YC S24 company that raised a $22M Series A from Spark and Matrix in 2026; the app is currently free

FAQ

What is Conductor?

Conductor is a Mac app that lets you run multiple Claude Code, Codex, or Cursor agents in parallel, each in an isolated git worktree.

How does Conductor work with Claude Code?

Conductor uses however you're already logged into Claude Code — API key, Claude Pro, or Claude Max.

Does Conductor use git worktrees?

Yes, each Conductor workspace is a new git worktree, keeping agents isolated.

Does Conductor compete with Tembo?

Partial overlap — both orchestrate coding agents. Conductor is local Mac-only; Tembo is cloud-based with enterprise integrations.

Executive Summary

Conductor is a Mac desktop application for running multiple Claude Code, Codex, and Cursor agents in parallel.[1] Using git worktrees for isolation, it enables concurrent agent work without merge conflicts. The app targets Mac developers who want parallelism without cloud dependencies, using their existing model subscriptions. Once an indie-looking side project, Conductor is now a venture-backed company: a YC Summer 2024 startup founded by Charlie Holtz and Jackson de Campos that announced a $22M Series A from Spark and Matrix in 2026.[2][3]

AttributeValue
CompanyConductor (YC S24)
Founded2024
Funding$22M Series A (Spark, Matrix; seed led by Matrix, plus Y Combinator)
Employees~6 (as of June 2026)
HeadquartersSan Francisco, CA

Product Overview

Conductor is a Mac desktop application for running multiple coding agents in parallel.[1] It creates isolated workspaces using git worktrees, letting you spin up Claude Code, Codex, or Cursor agents that work independently without stepping on each other's changes. Cursor support landed in v0.63.0 (June 2026) alongside a new Dispatcher feature.[4]

The app focuses on a simple workflow: add your repo, deploy agents to isolated workspaces, then review and merge their changes from a unified interface.

Key Capabilities

CapabilityDescription
Parallel AgentsRun multiple Claude Code, Codex, or Cursor instances simultaneously
Git WorktreesEach agent works in isolated worktree
Unified DashboardSee all agent status and pending reviews
DispatcherRoute tasks to agents (added v0.63.0, June 2026)
No Extra CostsUses existing Claude/Codex subscriptions (Cursor needs an API key)
Local-FirstAll code stays on your Mac

Product Surfaces / Editions

SurfaceDescriptionAvailability
Mac Desktop AppNative macOS applicationGA
Agent DashboardStatus and review interfaceGA

Technical Architecture

Platform: macOS only (native Mac app)

Isolation model: Git worktrees[5] — each agent workspace is a separate worktree, meaning:

  • Agents can work in parallel without merge conflicts
  • Changes are isolated until you explicitly merge
  • Same repo, different branches, concurrent work

Key Technical Details

AspectDetail
DeploymentLocal Mac app
Model(s)Claude Code, Codex (bundled, via existing subscriptions); Cursor (via API key)
IntegrationsGit
Open SourceNo (proprietary)

Agent integration: Conductor doesn't include its own model access. It bundles Claude Code[6] and Codex for version compatibility, using your existing login/API keys; Cursor integration requires a separate API key.[7] The app ships near-weekly releases — v0.64.0 landed June 9, 2026 with Claude Fable 5 support, days after Cursor support and Dispatcher in v0.63.0.[4]


Strengths

  • True parallelism — Multiple agents working simultaneously on the same repo
  • Free and no additional costs — The app is currently free and uses your existing Claude/Codex subscriptions[7]
  • Clean isolation — Git worktrees prevent agents from conflicting
  • Local-first — All code stays on your Mac
  • Simple UX — Dashboard shows agent status and pending reviews
  • Native performance — Mac-native app, not Electron
  • Rapid release cadence — Near-weekly updates through mid-2026 (v0.54.0 to v0.64.0 between mid-May and June 9, 2026)[4]
  • Funded runway — $22M Series A from Spark and Matrix reduces abandonment risk[2]

Cautions

  • Mac-only — No Windows, Linux, or cloud support
  • Limited agent support — Claude Code, Codex, and Cursor only; no open-source agents
  • No sandboxing — Agents run with your full user account permissions; the official FAQ suggests a dedicated machine or VM if that concerns you[7]
  • No enterprise features yet — No team management, no Jira integration, no signed commits; paid collaboration features are planned but not shipped[7]
  • Individual-focused — No collaboration features for teams
  • Monetization unproven — Free today with no published pricing; venture funding means paid tiers are coming, on terms not yet known
  • Limited integrations — No issue tracker support

What Developers Say

"Managing many agents, each in their own sandbox, felt like indisputably the future after using conductor for a day."

— an HN commenter whose team moved off Cursor[8]

"starting to feel clunky compared hacking directly against online live reloading apps"

— an HN commenter on where the workflow strains[9]

"reading the conductor docs it says all perms are given to agents by default"

— an HN user flagging the lack of sandboxing in a thread on agent guardrails[10]


Pricing & Licensing

TierPriceIncludes
StandardFree (as of June 2026)Full Mac app access

Licensing model: Free today; the team says it plans to charge for collaboration/team features in the future[7]

Hidden costs: Requires existing Claude Code or Codex subscriptions; Cursor agents require your own API key


Competitive Positioning

Direct Competitors

CompetitorDifferentiation
EmdashEmdash supports 20+ agents and issue integration; Conductor supports three (Claude Code, Codex, Cursor) with simpler UX
Codex AppCodex App is OpenAI-only with cloud sandboxes; Conductor is local with multi-provider
TemboTembo has enterprise features (Jira, signed commits, BYOK); Conductor is individual-focused
CrystalCrystal has A/B testing; Conductor has simpler UX

When to Choose Conductor Over Alternatives

  • Choose Conductor when: You want simple Mac-native parallelism for Claude Code, Codex, or Cursor
  • Choose Emdash when: You need broader agent support or issue tracker integration
  • Choose Codex App when: You want cloud sandboxes and OpenAI-native features
  • Choose Tembo when: You need enterprise integrations and team features

Ideal Customer Profile

Best fit:

  • Individual Mac developers who want parallel coding agents
  • Users with existing Claude Code or Codex subscriptions
  • Developers who prefer local tools over cloud services
  • Engineers who value simple, focused UX

Poor fit:

  • Teams needing collaboration features
  • Enterprises requiring Jira, signed commits, or compliance
  • Windows or Linux users
  • Developers wanting to use agents beyond Claude Code, Codex, and Cursor

Viability Assessment

FactorAssessment
Financial HealthStrong — $22M Series A from Spark and Matrix (2026), on top of a Matrix-led seed and Y Combinator (S24)[2]
Market PositionStrengthening — one of the better-known names in Mac agent orchestration
Innovation PaceHigh — near-weekly releases; Cursor support, Dispatcher, and same-week Claude Fable 5 support shipped in June 2026[4]
Community/EcosystemGrowing — individual developer tool with venture backing and YC network
Long-term OutlookFavorable — funded runway, though monetization is unproven and the category churns fast

The earlier read of Conductor as a two-person bootstrapped indie app is out of date. As of June 2026 it is a San Francisco company of roughly six people[3] with $22M in Series A capital and a fast release cadence. The open questions have shifted from survival to monetization: the app is free today, and the planned paid collaboration features have not shipped.


Bottom Line

Conductor validates market demand for agent orchestration UX. The git worktree approach is clever — true isolation without the overhead of multiple clones — and the $22M Series A from Spark and Matrix makes it one of the best-capitalized players in the Mac agent app category.[2]

However, the Mac-only, individual-developer focus still limits its market today. Teams needing enterprise features (Jira, signed commits, multi-user coordination) won't find them here yet, though paid collaboration features are on the roadmap.[7]

Recommended for: Solo Mac developers who want simple, free parallel Claude Code/Codex/Cursor execution with local-first architecture.

Not recommended for: Teams, enterprises, or anyone needing broader agent support, sandboxed execution, or issue tracker integrations.

Outlook: Materially improved since early 2026. With venture funding, a six-person team, and near-weekly releases, abandonment risk is low for now. The risks have shifted to monetization (free today, paid tiers undefined) and category churn as Codex App and other well-funded alternatives expand.


Research by Ry Walker Research • methodology