The best AI workflows create more time for the work only people can do.
That was driven home to me when I spoke with Cathy McPhillips, chief marketing officer at SmarterX at MAICON, the AI marketing conference. While we were talking, she had an AI agent running in the background. McPhillips also leads marketing across the Marketing AI Institute’s brands, including MAICON. So, she had the agent look for the top 100 marketing AI practitioners to prioritize outreach for an upcoming flash sale.
“It seems so basic,” she said. “But it was running while I was doing something else.”
That reflects how McPhillips and her team use AI throughout the event lifecycle. AI supports planning, content creation, and post-event marketing, while people remain responsible for customer communication, programming decisions, and speaker relationships. With event teams facing tighter budgets and higher expectations, this is a practical example of where AI adds value — and where it doesn’t.
The world’s most powerful SEO platform, purpose-built for Enterprise.
Pre-event: AI builds the context, people make the calls
Planning: Giving judgment its context back
The first work McPhillips’s team gave to AI was work no one likes to do: The competitive research of scouring websites, taking screenshots, and building spreadsheets of sponsors, packages, and logos. Agents can gather that and help the team prioritize which prospective sponsors to call first. Pricing gained more structure. Sometimes the analysis shows that a price increase makes sense, and sometimes it pushes back against it.
Forecasting presented a harder problem because MAICON has almost no normal history to work from. Each past year was a one-off that numbers alone couldn’t explain. Now she can query the event’s history like a colleague with perfect recall and ask whether a change represents a trend, an anomaly, or a forecasting mistake.
AI didn’t replace judgment. It recovered the institutional memory that judgment depends on. She uses that time to work on tasks that benefit from actual relationships. “All that manual work I can now invest in being the human, talking to my human connections.”
Promotion: The inbox stays human
McPhillips draws her clearest boundary around email, their most important channel. “People have chosen to let us enter their inbox,” she said. “I don’t know if I’m ready for a customer to receive something that doesn’t sound like us. I want to be their best stewards.”
One experiment showed why. A persona GPT the team built and trusted recommended email changes that seemed reasonable — until opens and clicks dropped. When she asked AI to diagnose the decline, it suggested the emails were no longer answering customers’ questions.
“You were the one who told me to do this!” she recalled thinking.
The experience reinforced a lesson that carries through every workflow: AI is not a shortcut around strategy. “We really need to do all of this with purpose and intent.”
Programming: Customer feedback designs the experience
The clearest example of AI serving attendees happens months before the event. The programming team compares a draft agenda with feedback from conference evaluations, Slack discussions, webinar chats, and podcast comments. That voice-of-customer knowledge helps the team identify where the agenda is strong and where it falls short. One review found a gap between attendees just beginning to use AI and advanced practitioners who felt they had outgrown the content.
That led to two new session formats. A transformation stage will feature 15-minute interviews with CMOs focused on one thing they have actually implemented. Build sessions will ask attendees to open their laptops and leave 30 minutes later with a working agent.
Not every improvement required AI. Chasing speakers for session titles and abstracts is a familiar challenge for event teams, so the team changed the process instead: speakers submit both with their signed agreement. No abstract, no contract.
During the event: A one-hour clip machine, plus speaker kits
Once the event begins, Goldcast connects to the main stage recordings, and within an hour of each session, the team has 10 short clips ready for social media. Attendees arrived the next morning, saying they missed the first day and wanted to catch up. By the end of the event, the team has about 150 clips to build anticipation for the following year.
The day after the event, every speaker receives a kit with highlights from their session, audience questions, clips, and graphics. It’s designed to recognize the speaker’s investment while providing them with material to promote their work and help them reach their next stage.
“We’re trying to create this experience for our speakers, who then [say], ‘I will come back and work with you anytime you need something, because you are also helping me.’”
The kits are assembled in a Claude project before a person reviews everything. “We have humans in the middle of everything,” McPhillips said.
Post-event: AI finds the threads across days of content
Event teams have long repurposed event sessions, but McPhillips believes many still overlook the bigger opportunity. Posting a few clips to LinkedIn or YouTube misses the chance to turn an event into year-round marketing. “It’s such a miss, because there’s so much content,” she said.
AI helps the team review three days of sessions to identify recurring themes and connections among speakers who never shared a stage. Those themes can be turned into podcasts, social content, Slack discussions, and webinars.
The team is also mapping that content to the questions people search for to support traditional SEO and answer engines. That work is still evolving. “We’re trying to figure it out just like everybody else.”
4 questions to ask before adding AI to an event workflow
If you’re deciding where AI belongs in your event workflows, ask these four questions:
1. Does it remove administrative work or replace a human relationship?
Building lists creates time for outreach. Automating the outreach itself may weaken relationships. McPhillips’s agent enriched her contact list, but the messages still came from her. “I know probably 60 or 70 of the people personally,” she said, and each one got a personal note.
2. Will someone review the output before a customer, speaker, or sponsor sees it?
Inaccurate content damages trust. Speaker kits move through Claude, but a person verifies every element before it goes out because a speaker who receives the wrong feedback won’t care how quickly it arrived.
3. Does this workflow expose customer data to the tool?
Establish data policies before you experiment, not after something goes wrong. SmarterX removes identifying details from every piece of feedback before it enters the voice-of-customer knowledge base.
“We’ll put anything into AI,” McPhillips said. “We’ll even put our P&Ls in there before we put in any customer data.”
4. Can a tool already in your stack get you most of the way there?
SmarterX used this year’s budget to hire staff rather than add more software. Any new tool has to answer a few simple questions: Does the team need it? Can it trust it? Will it work with HubSpot?
McPhillips starts with an even simpler one: “Can a tool you’re already paying for get you 80% of the way there?”
The answers to these questions will evolve as AI tools improve. Responsibility for relationships with customers, speakers, and sponsors won’t. That’s where people continue to add the most value.
The post 4 questions to ask before adding AI to an event workflow appeared first on MarTech.