Describe it once. We build the system that runs it.

A Pilot runs one automation. A System connects several so they run as one flow: the result of one becomes the trigger of the next. Describe what you want in a sentence and it builds the whole system for you, or chain your existing Pilots together yourself. It runs on the same engine, and you can build and run it from your phone.

For a lot of solo operators, yes, and then some. Zapier meters every task and makes you build one Zap at a time on a desktop. BizOps Pilot is one flat price with unlimited runs, connects your automations into whole Systems, and lets you build them from your phone.

Native connectors cover the tools solo operators actually live in: email, Slack, SMS, Notion, HubSpot, and a Klaviyo template builder that composes on-brand emails from your own brand kit. On top of that, Connect apps in one click reaches 250+ more (Gmail, Shopify, Stripe, Airtable, Discord, Telegram, and on) with no account or API key of your own to manage. For everything else there are three more universal doors: any HTTP API can be a step, every Pilot and System has its own webhook, and the MCP connector links any Model Context Protocol server so every tool it exposes becomes a step. Giant catalogs are mostly long tail you will never touch. The question that matters is whether YOUR stack connects, and it almost certainly does.

Yes. On the Connections page, Pro users can add a custom connector: give it a base URL and a few actions, each with a method and a path, and it becomes a tool you pick in any Pilot step just like a built-in. For an authenticated API you add its key separately, encrypted the same way every other key is and never shown again. A custom connector runs through the same safety net as everything else: a Test Flight skips anything that writes, and every request is blocked from reaching private or internal addresses.

MCP, the Model Context Protocol, is the standard way an AI agent talks to tools. It works in both directions here. Outward: connect any MCP server on the Connections page and every tool it exposes becomes a step you can drop into a Pilot, so you inherit a whole ecosystem instead of waiting for us to build a connector. Inward: our MCP bridge lets an assistant like Claude Desktop list, create, and run your Pilots and read their runs, which turns your automations into something an agent can actually operate. Both run on the flat price. For contrast, Zapier's help centre says each successful MCP tool call spends 2 tasks from your plan allowance (published 2026-06-04), so an agent calling tools in a loop draws down the meter as it works. API keys and the MCP bridge are Pro.

Free during beta, or go Pro now for a flat $39 per month for unlimited active Pilots and unlimited runs, or a one-time $249 for a Founding lifetime seat (capped at 300). That is the entire pricing page.

No. There are no tasks, no credits, and no meter of any kind. Runs are unlimited on every plan, free included: a Pilot can fire every 5 minutes forever and the bill does not move. Free caps one thing only, which is how many automations run AT ONCE: 3 Pilots and 1 System active at the same time. That is a ceiling on how many things run, never on how often they run. Paused Pilots do not count against it, and you can keep as many of those as you like.

Nothing is billed on volume and no automation is ever stopped for billing reasons, because runs are not counted. On free, the one cap you can reach is the activation cap: with 3 Pilots already running, activating the next one asks you to pause one first. Pause any Pilot to free the slot, or go Pro for unlimited active Pilots and Systems. Structural fair-use limits do exist, so that one account cannot starve another: webhook deliveries are limited per minute, scheduled Pilots run at most every 5 minutes, and a heavy burst queues against your own earlier work rather than other customers'. Those delay a delivery briefly. They never cancel an automation and they never cost you money. Section 3 of the terms says the same thing in full.

A flat $39 a month, and it lifts every activation cap: unlimited Pilots and unlimited Systems running at once, instead of 3 and 1. It adds API keys for the REST API and the MCP server, custom connectors for any API we have no built-in for, per-entity brains, and unlimited connected accounts per tool where free allows one. Runs are already unlimited on free, so Pro buys capability, never volume.

A one-time $249 for the Pro plan, capped at 300 seats. The terms say this plainly, so there are no surprises later. Lifetime means the lifetime of BizOps Pilot, not your lifetime: if we ever shut the service down, a Founding seat ends with it. We are a small company and you should weigh that before buying. It covers the Pro plan as it evolves, including features we add to Pro later, but it does not entitle you to a different, higher tier if we introduce one. It is tied to one account and is not transferable or resellable. The same fair-use limits apply, there is no run meter, and we will not invoice you for usage, ever. The AI itself still runs on your own key and is billed to you by that provider: a Founding seat covers the platform, not your model usage. Section 8 of the terms is the full text.

Two structural reasons. First, every step gets only the context it needs: there is no ever-growing chat thread being resent on every call, so cost cannot snowball the way it does in chat style tools. Second, every AI step has an effort setting: Smartest for judgment calls, Cheapest for things that run every hour, Balanced for everything else. AI runs on your own key at cost, with no markup and no credits to watch.

Yes. Any step setting accepts placeholders: {{trigger.email}} pulls a field from whatever started the run, {{step_2.output}} uses an earlier step's result, {{item.name}} points at the current item inside a loop, and {{vars.name}} inserts a value you saved once on your Account page, like a signature or support address. Templates ship with these aimed already, so the welcome email goes to the lead who actually filled the form.

Yes. Add a 'for each' step after any step that produces a list (orders, leads, rows, anything) and every step after it runs once per item: draft a note for THIS order, send it to THIS customer, then the next one. A failing item never takes down the rest, runs are capped so a huge list cannot run away, and the loop's result flows onward as a list.

Four ways: run it on demand, POST to its webhook URL from any service, run it on a schedule from every 5 minutes to weekly, or forward email through an inbound-parse service like SendGrid or Mailgun.

For real AI results, yes. This is a genuine setup step, not a perk, so it is worth being straight about it: you open an account with Anthropic, OpenAI, Google, or OpenRouter, paste the key into Connections once, and that provider bills you directly at their price. We add zero markup and never resell inference, which is why there are no AI credits to buy here and nothing to top up. A local model through something like Ollama or LM Studio works too, with a base URL and a model name instead of a key. If you connect nothing, AI steps still run rather than breaking your Pilot, but they return a clearly labelled simulated result instead of real judgment.

On the key alone, honestly, it is not. Zapier ships bring-your-own-key too: in the model-based AI pricing they published on 2026-06-15, connecting your own AI account is priced at 1x against a default tier of 3x. That is a real, documented discount at the market leader, and pretending otherwise would be a lie you could check in a minute. The difference is what happens after the key. There, your own key buys a cheaper multiplier on a meter that keeps counting every task. Here there is no multiplier and no meter, because we do not count runs at all. Bringing your own key is table stakes in 2026. Refusing to meter is the part almost nobody offers.

Connector credentials are encrypted with AES-256-GCM before they touch the database and are never shown again after saving. Sessions are httpOnly JWTs, passwords are scrypt hashed, and every route enforces per-user ownership.

Yes. Every Pilot exports as plain JSON with one click, and imports the same way. No secrets travel in the file and there is no lock-in.

Yes. If you would rather hand it off, our sister studio growthlayer.co builds your automation on this same platform, proves it on up to 50 of your own past examples, runs it, and ships one improvement every week. Your self-serve subscription stays exactly the same. See the work-with-us page for how it works.

A 30 minute call where you walk us through the job that eats your week. Your first automation runs within 48 hours, with approval gates on anything that sends or spends. Before it touches anything real, we run it against up to 50 of your own past examples and hand you the scorecard: what it handled right, what wants your eyes, where it made mistakes. Then every week we review the runs, teach it something new about your business, and ship one improvement.

The guarantee: if your first workflow does not earn its fee in 60 days, the next month is free.

$1,500 setup, then $1,000 a month. One workflow at a time, done properly. No credits, no usage tiers, no surprise bills.