Most of the media industry automated its buying a decade ago, but influencer marketing never followed. A campaign with three hundred creators still means searching profiles by hand, chasing drafts and building the report weeks later, on spreadsheets and WhatsApp.
We built Swavy as an AI-native influencer marketing platform, with an agent for each step of a campaign, from discovery to reporting. Three years in, we know what AI does better than people, and what we would still never hand over.
The first is vetting. A person can realistically review only a creator’s posts and headline numbers. An agent watches every video in their history, reads the captions, listens to the audio, checks the audience data and past brand deals and comes back with the exact clip that breaks a brand’s rules. For a client that invited international creators to visit, our agents flagged a video posted three years earlier in which one invitee had made offensive remarks about the host country. Few manual vetting processes go that far.
The second is finding what nobody tagged. A luxury brand hosted a major event and wanted its earned media value (EMV), across everything its guests posted, not only the posts that tagged the brand. From a guest list of 250 names, our agents followed more than 550 accounts, across Instagram, TikTok and Snapchat for 12 days, catching stories, carousels and videos with no tag, including one where the product was visible on a shelf in the background for under a second. The agents found 50 per cent more content than the client’s manual count, changing the EMV figure brand reported internally.
The first is the creator’s face and voice. Much of the industry’s AI debate is about synthetic influencers and machine-written scripts, and we think that is the wrong place to put it. AI drafts the brief and reviews the video, but the creator stays the author, because an audience that trusts a real person is the reason this channel works at all.
The second is the relationship with top talent. Agents handle day-to-day coordination across large creator pools, where most creators want a fast reply and fair rate. Celebrities and long-term ambassadors want someone they know, so senior team members handle those relationships directly.
The third is accountability. An AI agent can shortlist a creator, draft a message or review a video, but it cannot answer to a client when something goes wrong. So a person signs off on the creators, the outreach and the content, and we would keep that rule regardless.
The people did not disappear, but their jobs changed. On one large regional campaign, planners went from spending half their week on data and reporting to about 5 per cent, with the time going into strategy, creator relationships and the client.
Nearly every agency and platform now claims to use AI agents. Three questions reveal what it actually does: does it watch the content, or just the numbers around it? Can it show the evidence behind a decision? And when it is wrong, who catches it?
Then ask for a pilot: the pitch shows what the AI is designed to do, the pilot shows what it can be trusted to do. That is how our clients started, from global brands to government entities, one campaign and market at a time, each pilot earning the next. The creator’s voice, the relationships and the accountability stay with people. Those are the parts we would never hand over.
By Tarek Abboud is a co-founder of Swavy , an AI-native influencer…