Your AI team, hire and run your specialist agents

Ford is your AI co-founder, and he runs a team of specialist agents that do real work for your store. You talk to Ford and he delegates to the right specialist, or you work with each agent directly. Everything they produce comes back to you to review, nothing leaves the platform, changes your store, or reaches a customer without your say-so.

Meet the team

You hire only the specialists you need. Each one stays in its own lane:

Iris, Creative Strategist, ad and UGC concepts, scripts, and finished static ad images from your real products.
Mila, Customer Support, drafts replies to your support tickets, grounded in the customer's real orders and your policies.
Noor, Email & Retention, drafts email campaigns (and designs them on-brand) ready to review and save into Klaviyo.
Theo, Product & Listings, rewrites your product titles, descriptions and SEO, and can turn a supplier sheet into ready-to-paste listings.
Cleo, CRO & Conversion, reads your funnel and hands you a prioritised list of A/B test ideas with the rewritten copy to try.
Vera, Market Research, competitor and market briefs (with sources) and demand trends, so you know what to push next.

See them all on AI Team, laid out as an org chart with Ford on top. Agents you've hired show their recent output; the rest have a Meet card to hire them.

How to hire an agent

On AI Team, click Meet on the specialist you want.
Ford gives you a short pitch grounded in your real numbers, what this agent would do for your store specifically.
Set the agent's identity (its name and how it works), then its authority, what it's allowed to do on its own (more on that below).
If the agent uses an outside tool (for example, Iris can make images on your own image-generation account), you can connect it here, or skip and do it later.
Hire, and the agent suggests a sensible first task you can kick off with one click.

Giving an agent work

There are two ways to put the team to work:

Tell Ford, in any chat, say something like “get the team to research my market and turn the top angle into some ad concepts”. Ford picks the right specialist (or plans a few in sequence) and reports back.
Go to the agent directly, open its profile and use the Give a task box on the Outputtab, e.g. Theo: “rewrite the listings for my top 5 sellers”.

Finished work lands in your approval queuefor review, and stays on the agent's Output tab afterwards.

You decide what each agent can do

Every agent is draft-first: it prepares the work and you approve it. For each sensitive action you set an authority level:

Ask me first, it prepares the work and waits for your approval (the default for anything that matters).
Autonomous, it can do that one thing on its own, for lower-stakes steps you're happy to let run.
Never, off entirely.

Nothing leaves without you

Anything with a real-world effect, sending a reply to a customer, updating a product in your Shopify store, creating a campaign in Klaviyo, stays turned off until you explicitly switch it on, and even then goes through your approval queue first. Change any of this anytime on the agent's Authority tab.

Teaching your team, they learn from your feedback

Your agents work like new employees: give feedback and they remember it. Every lesson is saved to that agent's Memory and applied to all its future work, and each finished package tells you which of your lessons it applied.

Send it back, in the approval queue, click Send back with feedback, type what should change, and the agent redoes the same task, Round 2 lands back in the queue showing the feedback it's answering. Repeat until it's right.
Your edits teach too, when you edit copy before approving, the agent learns the pattern (you'll see “what they learned” with an undo).
Say what worked, use ＋ Note what worked next to Approve so good habits stick, not just corrections.
Just tell them, a correction in an agent's chat (“never use emoji in hooks”) is remembered on the spot, and you can tell Ford “tell Iris to stop using discount language” and he files it into her memory.

Everything an agent learns is visible on its Memory tab, edit or delete any lesson anytime. When the lessons pile up, a one-click Tidy upmerges them into a short list of standing rules (you review before anything changes), and a lesson that's really a brand-wide rule can be promoted to the Brand Brain so your whole team follows it.

Each agent's profile

Open any agent to find:

Chat, talk to that specialist directly about its area.
Identity & Authority, its name/personality and what it's allowed to do.
Memory, what it's learned and what you've told it (you can add or remove notes).
Output, its finished work, and the “Give a task” box.
Activity, a log of what it's been doing.

What it costs

Your plan includes a generous monthly allowancefor your AI team's work, with bigger plans (Captain → Crew → Fleet) getting more. A usage meter on AI Team shows how much of this month's allowance you've used, you never see a per-task price. Tools an agent runs on your own account (like image generation) are billed by that account, not us.

Give your team your brand so their work matches it: your visual brand kit and brand files. Then try making some ads with Iris.