A fraction of a team. The same software.

A Capacity Unit (CU) is one unit of platform work, metered transparently per run. A bug fix is 3 to 8 CU. A small feature is 15 to 40 CU. A medium feature with frontend, backend, and tests is 80 to 200 CU. CU consumption reflects model time, agent fan-out across multiple coordinated roles, and review iterations. Every CU is visible on the dashboard. Overage is published at $0.25/CU for Pro and $0.15/CU for Team. No hidden ratchet.

Yes, on managed plans. We absorb LLM costs and route each agent to the most efficient model for the job, so you don't pay frontier-model prices for routine work. On BYOK plans, you bring your own keys and pay your provider directly while we charge a flat platform fee for the orchestration. No per-token markup in either model.

Yes. Every paid plan has a BYOK LLM option. Pro BYOK is $39/mo ($29 on annual), Team BYOK is $299/mo ($239 on annual), Enterprise BYOK starts at $3,500/mo. You connect your own OpenAI, Anthropic, Azure OpenAI, AWS Bedrock, Google Vertex, GLM Coding Plan, or Fire Pass account. You pay your provider directly for tokens. We charge a flat platform fee for the orchestration, agents, and product owner. No CU caps on BYOK. The LLM bill is the natural meter. Note: this is separate from BYO cloud account, which is the default on every paid plan regardless of which LLM track you pick.

Yours. On every paid plan (managed LLM or BYOK), the platform deploys into your own AWS, GCP, or Azure account. You hold the cloud keys. We never run your production code on our infrastructure and we never see your production secrets. This is independent of which LLM track you pick. Even on the managed-LLM plan, the cloud account is yours.

We route across multiple frontier and cost-efficient providers, picking the best model for each task rather than running everything on one expensive model. We don't publish the exact routing because it changes as providers ship new models and shift pricing. What's fixed: no training on your code, no per-token markup, frontier-grade quality on the decisions that matter, and cheaper models on the routine work. Enterprise customers can pin model choice; BYOK customers bring their own keys and rates.

You can soft-overflow at the published rate. Pro $0.25/CU, Team $0.15/CU, Enterprise $0.05 to $0.10/CU. We never hard-stop without warning. You get an alert at 80% of your monthly allotment and another at 100%. Annual commits get 10 to 25% off overage rates. You can also set a hard spend cap at any threshold; new runs queue (default, resume next cycle) or block (stop immediately) based on your preference.

Yes. Every account has a configurable monthly spend cap. Set a number; below it, runs continue at overage rates; above it, new runs queue or block. Useful for predictable budgets, compliance-sensitive teams, and avoiding runaway loops in long-running agent sessions. The cap covers both managed CU and any human-in-the-loop hours billed in the same period.

Nothing you didn't choose. The biggest complaint about AI coding tools in 2026 is surprise bills: drained subscriptions, overages several times the monthly fee, quiet price changes. Our plans are flat monthly tiers with published overage rates, you get alerts at 80% and 100% of your bundle, and you can set a hard spend cap that queues or stops runs past a ceiling. An agent can't loop quietly and burn your budget.

No. Not on any plan, including Free. We don't train on your code or fine-tune on your data, and we hold zero-retention terms with the model providers we route to. This matters more in 2026 than it used to: some major tools moved to training on user code by default, leaving you to opt out. We do the opposite, contractually, at every tier.

No. Never. On any plan, including Free. No fine-tuning on customer data. Zero retention with the model providers we use. This is contractual at every tier, not a setting you have to enable. On managed plans, we route only to providers with no-training-by-default contracts (Anthropic, OpenAI Business and Enterprise tiers, Azure OpenAI, Bedrock, Vertex). On BYOK, you control the contract with your provider directly.

Nothing breaks. Free CU resets at the start of each month. If you exhaust your 100 CU and want to keep going, upgrade to Pro at $69/mo monthly or $59/mo annual. Your existing project, code, and configurations stay intact. Downgrading back to Free later preserves your work. You just lose new-run capacity beyond the free allotment.

Yes. Upgrades take effect immediately and prorate the difference. Downgrades take effect at the end of the current billing cycle, so you don't lose paid CU you've already committed to. CU allowances reset on the new cycle's start regardless of plan changes. Switching between Managed and BYOK is also possible mid-cycle, and we prorate the platform fee.

Yes. Pro is $59/mo on annual billing ($708/year) versus $69 monthly. Team is $449/mo on annual ($5,388/year) versus $549 monthly. That's about a 17 to 20% saving. Enterprise is custom-priced with annual as the default. Annual commits also include 10 to 25% off overage rates and priority support response.

A 5-person US engineering team costs $40K to $100K per month fully loaded. A hybrid US-lead-plus-offshore team runs $25K to $50K. An all-India or all-LATAM team is $12K to $45K. Our Team plan plus typical usage stays at $2.5K to $4.5K. That works out to between 3% and 12% of whichever team you'd hire instead. Pro plan plus typical usage stays under $200/mo for solo founders. That's less than one billable hour of a senior engineer.

Three common moments. (1) The agent quality gates escalate a complex review (security audit, performance regression, contract break). (2) You want a senior engineer to sanity-check a vertical slice before production. (3) Architecture decisions span more than one feature and need a human call. Four named roles are available: Product Owner, Senior Engineer, Architect, and Security & Compliance. Rates are quoted per engagement and flex with the staffing model. Both onshore (US and EU) and offshore (LATAM and India) options are available, and we will match the mix to the work and the timezone coverage you need. Team plans include 4 hrs/mo of PO time. Enterprise includes a 16 hrs/mo flex pool across any role. Block prepay and flex-pool arrangements both get a volume discount.

We offer a no-questions 14-day money-back guarantee on first-time Pro and Team subscriptions. After that, unused CU don't roll over and aren't refundable. Overage and prorated downgrades work normally. Enterprise refund terms are negotiated per contract. BYOK platform fees are also covered by the 14-day guarantee on first-time signup.

All major credit cards (Visa, Mastercard, Amex, Discover), ACH transfer for Team and Enterprise, and wire transfer for Enterprise annual commits. We invoice in USD by default. EUR, GBP, and AUD billing available on annual Team and Enterprise plans. Enterprise customers can also pay via PO and net-30 terms.