How Do You Sell a Personal AI Agent That Can Do Almost Anything?

An assistant that can do almost anything gives a founder an awkward marketing problem. You have to choose the one thing that makes a stranger want to try it.

Look at the category through a prospective customer’s eyes. Instinct describes an assistant that connects to your information, follows up on unfinished tasks, and works through text or calls. Poke offers help with email, meetings, and reminders through messaging apps. There are differences between these products, but a person scrolling past an ad may never get far enough to discover them.

“Your AI personal assistant” leaves that person with a lot to figure out. Why this one? What would I give it first? Is the benefit worth connecting my accounts?

This is where I’d use UGC. Several creators can put the same product into different recognizable situations, with different reasons to care. A founder gets more ways to reach customers and more opportunities to find a promise that customers actually respond to.

The appeal is speed. You can commission several ideas in parallel instead of spending weeks polishing one launch video. But the batch has to help you decide what to do next. Twenty videos that all say “this AI saves me hours” give you twenty chances at distribution and very little insight into why someone would choose your product.

Find the reason someone would try your agent first

If I were running acquisition for a personal-agent startup, I’d start with the customers who already get something useful from it. What did they first hand over? What had they been doing instead? What made them comfortable enough to let the agent help?

Imagine your best early users are independent consultants. One wants help catching client emails that need a reply. Another wants a useful briefing before each meeting. Both might describe the product as a personal assistant. Their reasons for paying could be quite different.

That difference matters when you commission content. A video about missed client replies can speak to the fear of appearing unreliable. A meeting-prep video can speak to the embarrassment of opening a call without remembering the last conversation. “Save time” is true of both, but it gives the creator much less to work with.

The next question is why your product is particularly good at that job. Perhaps it works inside the messaging app the consultant already uses. Perhaps it brings together information that otherwise takes several searches. Perhaps the user can check a proposed action before anything is sent. Use the advantage your product actually has; a competitor’s impressive demo is a poor reason to advertise a workflow you can barely support.

A creator who does this work can make the distinction visible. They know which client email is awkward to answer and what useful preparation looks like. An AI enthusiast might make a compelling discovery video, but you shouldn’t assume their audience has the same problem or willingness to pay. Those are different acquisition bets.

There is some encouraging evidence for specific demonstrations. Creator Kunwar’s Lindy case study reports 1.03 million views and 5,000 links sent across four reels, including a bakery phone agent and client outreach for creators. Lindy is a broader workflow product, so this is adjacent evidence. The report shows attention; it doesn’t disclose spend, paying customers, or retention. I’d borrow the specificity of the concepts, then measure the customer outcomes in my own campaign.

Give the first batch a question to answer

The first campaign doesn’t need to discover every possible audience. It needs to make the next spending decision easier.

For that consultant example, I’d consider a batch of 12 videos: three creators, each trying two supported workflows and making two openings for each. One opening could emphasize the problem: a client thread left unanswered. The other could emphasize the result: starting the morning with the replies that need attention already identified. The creators would use their own language and circumstances.

Twelve is an illustrative starting point, not a magic sample size. The useful part is that the batch has a structure. You can compare two openings from the same creator, see whether a workflow travels across creators, and decide which idea deserves another attempt. It’s much harder to learn from a dozen unrelated videos made for a dozen unrelated audiences.

Choose creators with credible access to that audience, then vary how they communicate. You might compare someone who teaches client management with someone who documents running a small consultancy. Check the subjects people respond to in their recent posts, and whether their viewers include potential customers. Follower count won’t tell you that on its own.

Give each creator the product before asking them to sell it. Pay for the time spent trying the workflow, including setup and feedback. A task that needs a founder on a call for an hour may make a great recording and a disappointing first experience for everyone who clicks.

The brief should settle the facts: what works today, what the customer must connect or provide, what requires approval, what result the video should show, and the disclosure and usage rights. Then leave room for the creator to explain why it matters. Lindy’s current creator program requires output from a real workspace; that’s a useful standard to adopt. You want someone’s experience of doing the job, not a convincing imitation of using the product.

You can move quickly without asking every creator to start from scratch. A shared product walkthrough, working example, and clear review deadline reduce repeated explanations. Track the audience, workflow, opening, format, and creator for each submission. A simple record is enough at first, provided you can connect it to the published post and campaign link later.

Follow the customer past the click

The first results will be tempting to read as a ranking. One video gets ten times the views of another. Surely that’s the winner.

It might be. It might also be reaching people who enjoy watching AI demonstrations and have no intention of paying for another subscription.

Give each post a trackable link where the platform allows it, and record its destination. For the client-email video, the destination should help a visitor try that workflow. Sending them to a blank chat box with “ask me anything” makes them invent a use case you just spent money explaining.

There’s a useful warning in Ohai’s relaunch announcement: people liked the idea of the assistant but didn’t always understand how to use it. That’s an acquisition problem as well as a product problem. An ad can create interest while leaving the visitor unsure what to do next.

For the consultant campaign, I’d follow visitors through connecting the required account, getting a useful first result, returning when another client thread or meeting comes around, and paying. Keep those outcomes associated with the creative that brought them in where attribution permits. Campaign links help, but they won’t capture every later visit or cross-device signup.

The differences tell you what to change. If people click but stop at account access, the video may need to explain what access is for, or the product may need a less demanding first task. If they finish the task but never return, a more exciting opening is unlikely to solve the whole problem. If one creator attracts fewer visitors who use and pay for the product, that creator may deserve more of the budget.

Set the spending limit against the economics of your subscription, including delivery and support costs. As a hypothetical example, $3,000 of total campaign spend producing 30 new paying customers is $100 per acquired customer. Producing 100 is $30. You still need to understand how long those customers stay. Neither calculation is improved by a million views on the campaign report.

Scale the idea across more creators

Once an idea looks promising, the next purchase should help establish whether it works beyond the original post.

Organic distribution is uneven. Two posts with different view counts aren’t a controlled test of their openings. Before committing a much larger budget, ask new creators with a similar audience to try the promising workflow and message. Where appropriate, licensed versions can also be tested through paid distribution with comparable audiences and budgets; agree on those rights before commissioning the content.

Keep the reason to care consistent while letting the circumstances change. One consultant can show the follow-up they owe after a proposal. Another can show a client question buried beneath routine notifications. Copying the original script word for word wastes the new creator’s knowledge of their audience.

This is how a useful batch can become a larger program. You have an idea to commission again, examples to brief from, and a customer result to check. Then you can test a new audience, another workflow, or a different promise without changing all three at once. Expand the part you have reason to believe in while continuing to look for the next one.

I’d also commission follow-ups from creators who keep using the product. What happened the next time the task came around? What became easier after setup? For a subscription, that can make a stronger argument than another demonstration of the first impressive result.

Make the next campaign easier to run

At some point, the founder’s constraint becomes getting the next batch out. Creators need answers. Submissions need review. Someone has to check the posts, handle payments, and keep enough records to understand what ran.

If all of that lives in the founder’s DMs, adding creators can make the program slower precisely when it should be getting faster.

Mainstage brings briefs, creator sourcing, submission review, verified views, payouts, and campaign reporting into one place. You can commission approved content, pay against verified views, or combine the two. Bring an existing roster, invite creators, or use applications and sourcing support to build a cohort. You’ll still need your product analytics to see which visitors become active, paying customers.

That gives the founder more time for the decisions that improve the campaign: which customer to pursue, which promise to test, what to change in the brief, and where the next dollar should go.

If you’re planning your first program, bring a target audience, one or two workflows your agent reliably handles, and the question you want the first batch to answer. That’s enough to start a useful conversation about creators, creative, and budget. Plan your first UGC campaign with Mainstage.