Frequently asked questions about Overclock
Starting with the question everyone asks on the livestream — then product, plans, compatibility, squads, and the technical side.
Pay for the models again? No.
Overclock connects the CLIs you already have — Claude Code, Codex, Gemini, Antigravity and more — using the login that's already on your machine. Each agent runs on the subscriptions and tokens you already pay for; Overclock doesn't charge anything for model usage. The Overclock subscription is just the cockpit: the grid, the maestro, the squads, and the 52 MCP tools.
What is Overclock?
Overclock is an ADE — Agentic Development Environment: the environment where AI agents, coordinated by a maestro (the orchestrator), ship software in parallel, in visible panes, with 52 native MCP tools.
I'm not technical — do I need to know how to code to use Overclock?
Not the traditional way. In Overclock you say what you want done — in text or voice — and the AI agents write, test, and ship the code; you command and review, the maestro executes. Knowing how to code isn't a requirement, it's an accelerator: people who know code go deeper, and people who don't learn by watching the agents work in visible panes, step by step. What you need to bring is clarity about the result you want. If this way of building is new to you, the full explanation lives at /vibe-coding.
What is a squad?
A ready-made team of AI agents coordinated by a maestro. You pick a squad by the result it delivers — a site, a video, a QA pass — give it a brief, and run it. The maestro splits the work, holds each agent accountable, and merges the results.
What are the 3 mission modes?
Free: you explore, commanding each pane directly. Squad: you give a brief and a maestro coordinates the team through delivery. Agentic: the unified mission with deep autonomy — exclusive to the Ultra plan.
What's the difference between Boost and Boost Pro?
Boost Pro is everything in Boost, plus squads: the arsenal's ready-made teams with recipes (Squad mode) — and ClockVoice, with push-to-talk right in the active pane and PT → EN → PT-BR. Boost has the cockpit, Free mode panes, agents, and skills.
What exactly am I subscribing to?
One subscription. The whole cockpit. Agents and skills included — and on Pro, ready-made squads, recipes, and voice command.
Can I cancel anytime?
Yes. The subscription is managed on Stripe: upgrade, downgrade, and cancel through the Stripe Portal, no need to talk to anyone.
Can I use an "unlocked" version of Overclock without subscribing?
It doesn't make sense — and the reason is the architecture, not a lock. What the subscription unlocks lives in your account: squads, agents, skills, recipes, and the constant updates (the product evolves live, week by week). And the models never came from the app: each agent runs on the subscriptions of YOUR OWN CLIs — Claude Code, Codex, Gemini — with the login on your machine. A loose copy is an empty cockpit: no account, no squads, no updates, and not a single extra token beyond what you already had.
Does it work with the Claude Code, Codex, and Gemini I already use?
No stack switch required: Claude Code, Codex, Gemini and more — the CLIs you already use, now side by side.
Which models can I use? Does Fable 5 work?
It works — the model comes from YOUR CLI, so whatever your CLI runs, the cockpit runs: Fable 5, Opus 4.8, and Sonnet through Claude Code; GPT-5.x and Codex Spark through Codex; Gemini 3.x through the Gemini CLI; MIMO, DeepSeek, and local models (Ollama, LM Studio) through the providers. When a new model ships in your CLI, it shows up in Overclock without waiting on an update from us.
I already have Claude Code on the Max plan — what does Overclock add?
Overclock doesn't replace your Claude Code: it multiplies it. Instead of one conversation at a time, your same login runs across several sessions side by side, each in a visible pane — and a maestro splits the request, holds each agent accountable, and merges the results, with token and cost on screen in real time. Your Max plan is still the fuel; Overclock is the cockpit that makes it run in parallel, with ready-made squads, skills, and recipes. The honest, criterion-by-criterion comparison lives at /comparacao.
Does it run on my system?
Mac, Windows, and Linux — with automatic app updates.
How do I run a squad?
In 3 steps: brief (you say what you want done, text or voice), squad (the maestro splits the work across the AIs, holds each one accountable, and merges the results), and delivery (you watch each pane, each delivery, and each cost in real time). The full walkthrough lives at /como-funciona; the showcase, at /squads.
How do two Claude Codes work on the same project without stepping on each other? Who's the brain?
Each agent runs in its own pane, and each mission can have an isolated git worktree — two jobs on the same project never mix branches or context. The brain is the maestro: it splits the request, holds each agent accountable, and merges the results, without executing anything directly. And above the maestro, you — commanding and watching everything in real time. The technical detail lives at /features/missions.
Do I need to set up API keys?
Not for the essentials. Login with Claude comes built in (built-in OAuth), and the other CLIs sign in with the authentication that's already on your machine — the pane spawns the CLI you already use, with the login you already have.
Does the app update itself?
Yes. Overclock checks for and installs updates automatically, on all three platforms.