The questions a serious buyer asks before paying.

• Does Consensus Logic use the OpenAI or Anthropic API?

  No. It uses the supported command-line tools that come with your existing Codex and Claude subscriptions. There is no separate Consensus Logic inference bill and no API key to fund.

• Do I need both subscriptions?

  To use both AIs, yes: supported access to both Codex and Claude. The exact supported plans and command-line prerequisites are listed on the Download page and kept current as the vendors change them.

• Does this give me extra Codex or Claude quota?

  No. Vendor usage limits still apply. Consensus Logic uses the access you already have. It does not create additional vendor allowance.

• Will Consensus Logic save quota?

  No fixed percentage is promised. The purpose of independent planning, small checked steps and review is to reduce blind repair loops and unnecessary repeated execution. Actual vendor usage depends on the project and on your subscriptions.

• Why not just use Codex or Claude directly?

  Use them directly when a single model and a simple task are enough. Consensus Logic is for work where you want two independent opinions, your own say over what runs, small checked steps and a record of what happened, rather than one model planning, changing and grading its own work.

• Won't the next generation of models make this unnecessary?

  Better models improve the work Consensus Logic can coordinate. They do not remove the value of independent review, explicit authority and evidence. A more capable model can still misunderstand a system, optimise around the wrong assumption, or report success before the real application works. What matters is who holds authority and what counts as proof, and that does not change with a model release.

• Does it send my source code to Consensus Logic?

  The product is local-first: projects, plans, records and reports stay on your machine. The AI providers you have chosen receive project context under their own terms, because they are the models doing the work. Consensus Logic does not claim that no source ever leaves the machine when the provider workflow requires it. The privacy page describes source handling exactly as implemented.

• Is it a coding agent?

  No. Codex and Claude write the code. Consensus Logic decides what they are allowed to do, checks what they actually did, and keeps the record.

• Can I walk away and let it run all night?

  Not as an unlimited autonomous agent. Automatic supervision, where supported, still operates inside the permissions and budget you set, and its stopping conditions. There is no unattended scheduler, and one task executes at a time.

• What happens if the two AIs disagree?

  A real disagreement is surfaced to you with the trade-off stated. Consensus is not manufactured by weakening one side until it agrees. You decide, and the decision is recorded.

• What happens when a step fails?

  The controller records the failure and follows the repair policy you chose: the failure comes to you first, or a limited number of corrections run within the recorded budget. A failure is not silently converted into success because a model says it is fixed.

• What platforms are supported?

  Windows is the launch platform: Windows 11, x64. macOS and Linux are not supported. If that changes, it will be because enough people asked on the platform waitlist.

• Can it work offline?

  The application and its records are local-first, and the licence works offline after activation. The AI providers themselves need whatever connectivity their command-line tools need.

• What am I paying for if I already pay for two AI subscriptions?

  The layer around the AIs: recovery, two independent plans, review, limits the AIs cannot cross, small checked steps, project memory, a record of what happened, and the ability to pick up where you left off. The value is avoiding the uncontrolled repair spiral, not buying more model tokens.

• Why is Pro USD 399 per year?

  Because the relevant comparison is not another twenty-dollar utility. It is the cost of losing professional hours or days to repeated repairs and uncertain changes. Compare the price with your own last failed loop. The pricing page has the plans.

• What is Consensus Logic not?

  Not a coding agent, not an always-on agent, not a pay-per-use AI wrapper, not cloud-hosted, not a team product yet, and not available for macOS or Linux. Each of those is stated on the page where it matters, so nobody buys the wrong thing.

Vendor plans, limits and command-line tools change. These answers are updated when they do.