Building an AI Ecosystem to Supercharge Event Content
▼ Summary
– Over half of internet content is now AI-generated, marking a rapid shift in content creation.
– The article challenges the notion that AI content is inherently low quality, noting that both humans and AI can produce valuable or poor content.
– MarTech used an AI-powered ecosystem to efficiently atomize and amplify content from their September 2025 conference, with human oversight at every stage.
– Key steps included using Descript for transcripts and video clips, ChatGPT for insights and summaries, and Gemini for session evaluations.
– The process sped up content distribution and established a playbook for future events, with plans to integrate AI outputs into event management platforms.
Nearly half of all online material is now produced by artificial intelligence, a remarkable shift in how information is created and shared. For those who manage event content, this presents a powerful opportunity to enhance both the speed and scope of their output. Building an AI ecosystem to supercharge event content allows teams to transform lengthy recordings into a wide array of valuable assets, reaching audiences more effectively than ever before.
A common misconception suggests that content is either low-quality “AI slop” or genuine human art. In reality, the quality of any content depends entirely on the process behind it. Humans have produced mediocre work for ages, and AI is fully capable of generating excellent, useful material when guided properly. The key lies in a collaborative workflow where human oversight ensures quality at every stage.
At the recent MarTech Conference, we implemented such a system to expand the reach of our event materials. Every piece of content passed through human review before publication, but the heavy lifting of processing and repurposing was handled by AI. This approach dramatically accelerated our workflow, allowing us to deliver insights to our audience much faster.
The September MarTech Conference was a free, online event featuring six live panel discussions focused on data and artificial intelligence. Each session included a moderator and several participants, streaming directly to registered viewers.
The first step in breaking down these discussions involved converting the raw video into a format other tools could easily process. Our team used Descript, an AI-powered video editor, to create dozens of short clips from the full sessions. More importantly, the tool generated precise transcripts that identified each speaker automatically. By the end of the conference day, we had compiled nearly one hundred pages of transcriptions in a single Google Doc, forming the essential foundation for all subsequent content creation.
These complete transcripts were exported as a PDF and uploaded to ChatGPT. We instructed the model to scan the discussions and pinpoint the most compelling insights and quotes from the panelists. A human editor then reviewed this output, selecting the top items to feature on the MarTech website just days after the event concluded. We also formatted the best quotes into a social media carousel for additional promotion. Furthermore, ChatGPT produced draft summary articles for each session, which our editors refined and published in the following days.
To provide an alternative way to consume the event’s key takeaways, we used Google’s NotebookLM to generate an AI-powered audio overview. This twelve-minute summary was embedded directly within the published session recap articles.
Live conferences always carry an element of unpredictability. Although each panel had a preparatory call, the actual discussions were unscripted. To quickly evaluate each session’s effectiveness for our events team, we built a custom Gemini GEM. Using the session transcripts, Gemini compared each discussion’s actual content against its original title and description from the agenda. This analysis measured how well the session delivered on its promised topic. The tool also provided a detailed share-of-voice breakdown, calculating the speaking time for each panelist in minutes, seconds, and as a percentage of the total session.
Our experience with the September conference provided a solid blueprint for deconstructing and amplifying event content. While we successfully met our initial goals, we see even greater potential ahead. One area we intend to develop is linking our AI-generated outputs, such as session transcripts and performance evaluations, directly to our back-end event management platform. This system houses years of historical data from past MarTech and SMX events, including every session, speaker, and topic. Integrating new AI-derived insights will give our event planners a richer, more informed foundation for designing future conferences.
(Source: MarTech)