salesascode
← All feature docs
Feature docs · /onboarding

Getting started

Three steps from purchase to your first scheduled run: add your AI provider key, connect your data, choose where output goes. Your first nine AI runs are on us, so you can start before step one. One page for every kind of purchase. Nothing to install.

Page layout

The setup steps sit beside the nine-role team preview. Your existing business brief, provider-key, connection, and runtime controls stay in the setup flow.

From setup to the working day

Confirm your business brief, then connect the sources and credentials your chosen work needs. The existing setup controls save your progress. Labelled samples help explain the view before your own records arrive. Once a task saves real research or a draft, open it from Today or Work. Activity records completed and failed runs so you can check what happened.

Overview

Every "on cron" role works the same way. It runs the work for you on a schedule, on our servers. There is no desktop app, plugin, or marketplace to install. Whatever you bought, checkout lands you on Get started, which names your purchase, waits for the licence if it is still landing, and walks the three steps. You can return to it any time.

There used to be a second page for this, one per product at /get-started/<product>, describing four steps in a different order. Its copy drifted from the real flow, so it is gone and that address now resolves here.

Coming back from checkout

The page reads your actual licence rows before it says anything about them. Getting through Stripe is not the same event as the licence being written; the two arrive separately, seconds apart, on a callback.

Every kind of purchase gets this, not just a single role: All-Access, the Founder plan, Team on Cron, one role, or a one-time cart. Where a purchase has a next job of its own the page carries it through, so a team owner gets a link to invite their team and a cart buyer a link to their ledger.

Before the three steps

A new account's first screen is Home, with the sample standup already on it and one action at the top of the setup rail: your website. Your roles read what it tells them before they write anything, so a brief is about your market rather than a generic one. It needs no key and no CRM, which is why it comes first; reading the site runs on your seat, and the flow says so and sells the seat in place if you have none yet. If adding the seat answers "Checkout is not available right now", nothing was charged and we have already been told; what you typed stays on the page. If you skipped it, the link back is at the top of Get started.

The three steps

  1. Add your AI provider key. Your agents run on your key, so inference is billed to you by your provider and never by us, and it counts against no cap of ours. We do record each run in your usage log, with the model and the token counts, so you can see where your key is being spent. We store it encrypted and never read it back to the browser. One key covers every role the account holds. The free deterministic checks run without a key, so the schedule is not idle while you fetch one. Your first nine AI runs are on us as well, so you can watch a pass produce a draft before you go and find a key. When they are spent the console says so plainly rather than telling you to add a key you may already have added.
  2. Connect your data. One click per tool: Salesforce, HubSpot, Gmail, Google Calendar. We custody each token encrypted and read it on the schedule. Opt-in and reversible. You can skip this, but the deterministic checks read your CRM, so they start once a source is connected and not before.
  3. Choose where output goes. The in-app draft queue is on by default and needs no setup. Drive delivery writes each report into your Google Drive as a Doc plus a raw copy, per role. Gmail drafts and Outlook drafts can be chosen on Home once that mailbox is connected. Gmail drafts places your morning queue's outreach drafts in your Drafts folder, addressed and unsent. Outlook drafts are placed one at a time, after you approve each message in Activity. Placing every role's drafts in Gmail is built and switched off, because it needs a Google security review we have not started; the setup step says which applies to you.

Power users can also drive the account from their own scripts or an MCP client. That path is optional and lives under Settings, Advanced, see API Keys & MCP.

Your first import

Connecting a CRM queues an import of your accounts and open deals; a worker performs it, so there is nothing to press and nothing to wait on the page for. Get started shows a card telling you where that import got to: queued, running, finished with the counts, or failed. Before this the page said nothing either way, so a queued import and a failed one looked identical.

Once it has finished, Refresh the hints on those deals re-runs the second half over the deals it imported: the MEDDPICC scaffold on each, and the hints drawn from the patterns across your closed deals. It only fills gaps, so anything you have typed into a deal yourself is left alone, and running it twice when there is nothing to add says exactly that rather than going quiet.

When research stops

Research pauses with a sentence saying why, and the page offers the step that fixes it. A missing Anthropic key or an exhausted AI budget links to Settings. A launch that used all 30 of its AI calls cannot resume, so the page offers Review saved results instead. A busy AI service or the per-minute limit clears on its own; Resume stays available.

If your connection drops or the server is slow, the page says so in plain words and your work stays saved. A read is tried once more before it gives up, and a background research step waits longer between retries, up to a minute, while the service is down.

A contact that is not verified

Your first launch places a draft only for a contact whose address is known to arrive. A card marked Review before sending says why: only the address shape was checked, the domain accepts mail to any address, or the employer has gone stale. Its Approve and place in Gmail button stays disabled until that changes.

Verify contact asks your connected Apollo account about that exact address and records its answer on the card. A check uses one of your five launch credits. Checking the same address again within the reuse window costs nothing. An address Apollo cannot confirm stays held, so choose another contact at that account. There is no way to place an unverified address anyway.

Verifying it worked

Your first runs and drafts appear on Home and on Activity. If nothing lands, the two things to check are on Get started itself: whether a data source is connected under step two, and whether the runtime is on in your role's console. Home's standup names each role's next scheduled pass, so a role that has not run yet says so rather than looking empty.

See also Licensing & renewal for how yearly subscriptions, renewals and lapses work, and Your roles for the logged-in hub showing every role's status at once.