Keeping six clients apart inside one automation tool

Before you weigh any tier, decide what the spend is meant to buy. For a small agency, the value sits in doing connection work once and then asking the AI to act inside each client's own accounts, so sending and logging stay apart. The lifetime tiers on softwareprogrammer.app are the only published prices, and you should read them there rather than assume a figure.

Think about where your hours disappear now. If someone rebuilds the same outreach steps for every client each month, a single connection screen for the apps you already pay for removes that repetition. The activity log then gives you a record to point at when a client asks where a lead came from, which is the second half of your goal.

Your objection about one shared inbox is the right thing to test first. Because work runs inside your accounts, separation depends on how you set those accounts up, not on a shared pool. Ask yourself whether you can give each client a distinct sending account before you commit, and plan that structure in advance.

Two things split the tiers: volume and whether you bring your own model key. The published tiers are Starter lifetime, Growth lifetime and BYOK unlimited lifetime, and they differ along those lines. For an agency running outreach for six or seven clients, volume is the number that decides whether a cheaper tier fits or frustrates you mid-campaign.

The BYOK option changes who bills you for model usage. On that tier your own key is charged by your provider, separately from the lifetime tier you bought, which suits an agency that already holds a model account and wants to control that side. If you would rather not manage a key, the volume tiers keep model costs inside the platform arrangement.

Read the tier table on softwareprogrammer.app against your actual monthly sending, not a guess. Count the clients, the campaigns per client and the research steps each one needs, then match that to the volume each tier allows. The dividing line is rarely about features and almost always about how much work you push through in a month.

Access is the first check. The platform can only act within the permissions your connected accounts already grant, so an app locked down by a client's admin will limit what the automation can do. Confirm which accounts you control and which need approval before you plan a workflow around them.

Second, look at how destructive steps behave. Any step that would delete or overwrite data waits for your confirmation rather than running unattended, which protects client records but adds a pause to sequences. Build that pause into your expectations for anything that edits or removes entries.

Third, confirm the separation you need. Whether each client's sending stays distinct depends on the accounts you connect and how you organise them, so sketch that layout first. Also check the guided setup and activity log, since the log is what lets you answer a client's question about a lead's origin with evidence rather than memory.

Moving from an older setup takes planning rather than a single switch. List the apps each client's workflow touches, note the permissions on each, and decide the order you will connect them. The one connection screen lets you do this app by app, so you can bring one client live before touching the next.

Once a client is connected, test a small sequence end to end: research, write, send, update and log. Watch the activity log to confirm the steps landed where you expected and that the lead trail is readable. Do this per client before scaling volume, because a fault found on one account is cheaper to fix than the same fault across seven.

Then decide your tier with evidence. If your test runs sit comfortably within a volume tier, stay there. If you already hold a model key and want usage billed to it, the BYOK unlimited lifetime tier is the route. Either way, the upgrade is a sequence of small confirmed runs, not one large leap.