If an AI agent can turn an advertiser’s RFP into a proposal in minutes, what’s left for a publisher’s ad sales team to do?
A lot more selling, said Mark Howard, TIME’s chief operating officer.
In the recent past, when an RFP would come in from a potential buyer, it would kick off a flurry of activity – and a series of labor-intensive tasks – that had little to do with actual selling. Once a rep logged the RFP details in Salesforce, the hand-offs would begin.
Alerts would fire off via Slack, product experts would be dragged into Google Meets to map out inventory across a large product suite and notes would get blasted over email. Meanwhile, the sales team would spend days manually coordinating across various platforms to stitch together a response.
“Premium media sales is still a highly personal, relationship-driven business,” Howard said, “and sellers lose too much time playing coordinator.”
TIME well spent
To help ease that operational bottleneck, TIME has been working with Kana, a startup founded by Tom Chavez and Vivek Vaidya, serial ad tech entrepreneurs and the dudes who co-founded Krux, the DMP Salesforce bought roughly a decade ago.
In 2019, the pair launched super{set}, a nontraditional VC firm that co-founds, builds and incubates data and AI startups.
Along with Kana, its portfolio includes privacy platform Ketch; data clean room provider Habu, which LiveRamp bought for $200 million in 2024; and Rembrand, an AI-powered virtual product placement company that agreed to merge with content ad platform Whisper World just last week.
Kana, which came out of stealth in February with a $15 million seed funding round, builds AI agents that take over the grunt work of the ad business.
TIME is using two of Kana’s agents. One works alongside the sales team, reading RFPs and drafting proposals. The other works with ad ops, watching live campaigns and recommending what to do when one starts to fall behind.
The sales agent draws on what Vaidya refers to as a “knowledge graph,” a shared pool of context Kana builds by pulling from TIME’s Salesforce CRM, order management system and ad server, both direct and programmatic.
It also brings in data from live campaigns, which is atypical for a pre-sales tool, Vaidya said. Once a deal closes, this type of data is usually the domain of ad ops, which uses it to look back at how a campaign performed, not to plan the next overture.
The sales agent can proactively turn that information into raw material for pitches.
“We do the research, we look at historical data and we look at similar advertisers,” Vaidya said. “Then a salesperson can go in and start ideating on new proposals for advertisers that TIME hasn’t even sold to yet.”
But the relationship itself still belongs to the salesperson, Howard said, and TIME isn’t looking to automate that or make it into an agentic process.
“We don’t do it with our journalism,” he said, “and we’re not going to do it with our sales.”
Well, at least, “that isn’t the plan for now,” he added.
Ops gets an assist
Once a deal closes, the work shifts to ad ops, and to Kana’s second agent.
Ad ops teams already have dashboards that show them when a campaign is underdelivering or not pacing properly. What those dashboards don’t say, though, is what to do about it, Vaidya said.
The Kana agent answers that for each campaign and recommends a fix, like changing trafficking, pacing or budget, so a shortfall doesn’t turn into a make-good.
TIME’s revenue operations team, meanwhile, is also using the tool, Howard said, and he’s seen “a significant increase in what they’re able to accomplish in any given day or week.” It’s also helping with budget planning.
“And not just for how we’re going to finish this year and how Q1 looks, but actual 2027 modeling,” Howard said. “It’s a whole new world when you can have these kinds of insights delivered to you.”
Human relationships first
Between proposals and pacing, agents now cover most of a deal’s life cycle, so there’s technically nothing stopping a publisher’s agent from dealing directly with an advertiser’s agent while the sales rep heads out for a three-martini lunch.
Kana could build that, Vaidya said. Put an MCP server (a standard way for AI agents to connect to other software) in front of the proposal agent, and it would be able to talk to an advertiser’s agents tomorrow.
“That’s a technology problem, which is easy to solve,” he said.
But TIME isn’t planning to go there.
Selling is still built on relationships, Howard said, and agents doing the deal is “just not in our game plan.”
