One platform, priced from one person to the enterprise.

Beyond the usage your plan includes, AI usage is metered per call and billed at what the provider charges for it, plus your plan's commission on top. That commission pays for the platform around the call - the routing, the controls and the usage record - and it is 20% on Free, 10% on Starter, 8% on Professional, 5% on Business, and negotiable down to 0% on Enterprise. It applies to usage beyond what your plan includes. The price of a model is shown in the workspace before you send, so nobody is guessing.

The subscription covers seats, workflows, the platform itself and an amount of AI usage every month - $10 on Starter, $50 on Professional, $100 on Business. Usage beyond that is metered as you use it and billed at provider rate plus commission, so there is nothing to buy in advance. To see the subscription, the usage and the internal time side by side for your headcount, use the TCO calculator linked below the plans.

Every new workspace starts on Free, with a one-time $5 of AI usage on any model and no card. It is spent exactly the way paid usage is, at the provider's rate, so what you see on Free is what the work will actually cost. When you need more people, workflows or a monthly usage allowance, move up a plan - the workspace, the prompts and the documents come with you.

No. Each paid plan includes an amount of AI usage every month - $10 on Starter, $50 on Professional, $100 on Business, and up to $1,000 on Enterprise - and it resets at the start of the next month. It is the same on monthly and annual terms. The Free plan's $5 is granted once.

All of them, on every plan including Free. That covers models from OpenAI, Anthropic, Google, xAI and others, as well as open-weight families such as Llama, DeepSeek, Mistral and Qwen. You can pick the model per message, pin one for a workflow, or let the workspace pick the best model for the job. New models arrive without a contract renegotiation.

A workflow runs an agent with fixed instructions on a schedule - hourly, daily, weekly, monthly or a custom cron - or whenever someone presses Run now. It can use your connected integrations and MCP tools, and keeps every run in its history. Ordinary chats, documents and one-off prompts are not counted against it. Free has none, Starter includes 10, Professional 20, and Business and Enterprise are unlimited.

On Professional and Business, yes - extra seats are $5 per seat per month on top of the plan, with no cap and no renegotiation. Free and Starter are fixed at 1 and 3 seats, so growing past those means moving up a plan. Enterprise seat counts are set in the contract.

Yes, and it takes effect straight away. From an annual plan to a higher annual plan you pay only the difference for the rest of the term, and your renewal date stays the same. From monthly to any annual plan, the unused part of the current month is deducted from the annual fee. From monthly to a higher monthly plan, the difference for the rest of the month is charged at once.

Anything that lowers the commitment waits for the end of the period you have paid for. On an annual term, a lower plan or a move to monthly billing starts on the anniversary. On a monthly term, a lower plan applies from the next invoice. Any other change goes through a support ticket and we handle it by hand.

BYOK means bringing your own provider keys: from Business upwards you can point StickyPrompts at the OpenAI, Anthropic or Google accounts you already hold, and those calls are billed to your contract rather than ours. BYOM means bringing your own model - a fine-tuned or self-hosted open-weight model connected to the workspace and governed like any other. BYOM is an add-on at $99 per model per month on Business, and included on Enterprise.

EU-only endpoints so inference stays inside the region, the records and model documentation you need to evidence AI Act obligations, model policies and guardrails that decide what may be sent where, detection for sensitive information before it leaves the workspace, prompt injection detection (in development), and a privacy log of every guardrail trigger and policy change. They are controls and evidence for your own compliance work, not a claim that a product makes an organisation compliant.

That is what the enterprise options below cover: your own provider keys on dedicated infrastructure, privately hosted open-weight models with no third-party inference at all, or a full self-hosted install in your own data centre. Pricing there is a flat fee based on compute and term, and the usage commission is negotiable down to zero.