Using AI to Build a Thought Leadership Event in 7 weeks: ReCo LIVE 2025

Why we built it

ReCo LIVE 2025 sets the tone for ReCo's next chapter. We wanted to be known for what we know about audience insight, brand building and content with AI, so we stopped writing about it and started teaching it in person.

That meant treating the day as content rather than as an event. Everything we would normally publish — research, points of view, workshops, teaching — we put into a room and let people take part in it. This article is about how we produced all of that in 7 weeks, and how much of it AI actually did.

• Gather 100+ purpose-led founders, marketers and creative leaders in one room

• Share genuinely current knowledge on audience insight, brand building and content with AI

• Build the authority to be invited into those conversations, rather than buying attention

• Prove the content system we sell by running our own launch on it

01 We researched the audience before we planned a single session

We did not start with a run sheet. We started by working out who this room was for and what they were already worried about. That work ran for 9 months before the 7-week sprint: market sentiment, audience mindset, and an honest read of what our resources could carry.

AI did the heavy lifting. We used it to sweep the field — what purpose-led leaders in Sydney were publishing, asking and arguing about — then to cluster that into themes we could actually programme against. What would have taken an analyst weeks took days, which is the only reason a 7-week sprint was survivable at all.

The themes that came out of it became the programme: trust in the AI age, going beyond the brand, and the technical edge.

02 We brought Sydney's changemakers into one room

Knowledge sharing needs people worth learning from. We approached speakers and mentors on values alignment rather than audience size, and built the programme around what each could genuinely teach.

20+ speakers and mentors took part, from Canva, Dovetail, the City of Sydney, Decidr, PLAI, Tech Ready Women and Digital Storytellers. The room they drew was made up of leading universities, government organisations and well-known purpose-led brands — the people we wanted in the conversation, rather than whoever a paid campaign could reach.

03 We mapped the channels that would actually reach them

We built a channel roadmap rather than a campaign. Each channel got a job, a test and a number:

• Organic social to carry the point of view and the speaker announcements

• Paid social on Meta and LinkedIn, run as a test to find the audience profile rather than to fill the room

• LinkedIn outreach built on Sales Navigator lists and sequenced through Dripify

• Partner channels, with every partner given a comms pack so they could amplify in their own voice

• Direct invitation through existing client and community relationships

We read cost-per-acquisition across the paid tests to identify who was responding, then moved the weight onto the channels where trust already existed. Ads told us who to talk to. Relationships got them in the room.

04 We ran research, writing and production through AI

This is the part most useful to borrow. We built a custom brand GPT and ran everything through it, so 14 people producing content in parallel still sounded like one brand — speaker announcements, workshop outlines, ad copy, web copy, partner comms and post-event content.

Behind it we ran agentic AI on Relevance AI and n8n, with Zapier joining the pieces. Those systems handled the work that scales badly with people: research sweeps, list building and verification, drafting, and moving content between the tools that needed it.

The point is not that AI wrote the event. It is that AI absorbed the repeatable work so a small team could spend its time on judgement — who to invite, what to teach, what to cut.

05 We built one content system to feed every channel

One system produced everything: the website, the ticketing flow, organic social, paid creative, partner assets, the opening film, live streaming and same-day social content. The same system we build for clients is the one we launched ourselves on.

• Weeks 1–3. We locked the theme, approached speakers, mentors and partners, wrote the campaign messaging, and designed and built the web page and social content system.

• Weeks 4–5. We shipped the media kit, tested every channel, and mapped the customer profile against channel efficiency using live conversion data.

• Weeks 6–7. We produced presentations, the opening film and the experience design, and ran communication with speakers, mentors, partners, guests and the live-stream audience.

• Week 8. We ran the day from 6am, capturing photography, live stream and social content as it happened.

Because it was one system rather than a set of one-off deliverables, the content kept working after the day. The recap film, the replay and the talks are still doing the job the event was built to do.

06 We staffed it with students and specialists

We produced ReCo LIVE with a team assembled through our work-learning programme with UNSW and the University of Sydney. Students worked beside our strategists across strategy, design, marketing, content and production, most of them in paid roles.

• Danling Xiao — chief strategist, curator, creative director, partner communication

• Mary Walker — content management and production, partner engagement

• Noah Kirby (USyd, University of York) — early-stage planning

• Anett Petrovics — event management · Emran — web development

• Mayu Ashizawa (UNSW) — event branding design

• Riya Mathew (UNSW) — presentation, advertising and web page design

• Karen Oyama (USyd), Akshata Iyer (UNSW), Xitong Shao (UNSW), Faiz Zahirul (UNSW) — video content

• Gabrielle dela Cruz (USyd, Cornell) — research and content development

• Ellen Zheng (USyd, Boston University) — AI workshop development

• Mert Kayabas (USyd, UPenn) — LinkedIn outreach and list building

• Nina Jung — ad placement · Jason de Plater — photography · Ceremony Cast — live streaming

Several students received formal commendations from their faculties. 7 volunteers from The Impact Collab completed the crew.

07 The tools behind it

• AI: custom brand GPT, agentic AI on Relevance AI and n8n, Zapier

• Design and content: Figma, Canva, Notion

• Web and ticketing: Webflow, Humanitix

• Data: Google Analytics, Google Tag Manager

• Outreach: Meta ads, LinkedIn ads, LinkedIn Sales Navigator, Dripify

Project impact

• 250+ people involved — 100+ in the room, 100 on the live stream, plus speakers, mentors, partners, crew and volunteers

• 20+ speakers and mentors and 10 partner organisations

• 7 weeks from concept to doors open

• A recap film, a full replay and a body of talks that keep working long after the day

“The event was well planned, thoughtfully curated, and smoothly executed. I was impressed by the quality of attendees.” — Diana Dong, Managing Director, Sussex Australasia

“Attendees weren't just learning. They were connecting, sharing, and building ideas together.” — Sarah Macaulay, FD Global Connections

Key takeaways

• Research first, programme second. AI compresses audience research from weeks to days, which is what makes a 7-week build possible.

• Teach, don't promote. The fastest way to build authority is to give away what you know, in public, alongside people worth learning from.

• Give every channel a job. Paid finds the profile; relationships fill the room. Test to learn, then move the weight to where trust already sits.

• Run one content system, not a campaign. A system keeps producing after the launch date; a campaign stops.

• Use AI for the repeatable work. It buys back the hours a small team needs for judgement.

FAQs

How do you build thought leadership quickly?

By publishing what you know rather than promoting what you sell, and by doing it alongside people worth learning from. We spent 9 months on audience research and relationships, then 7 weeks producing. The speed comes from knowing exactly who you are talking to before you start making anything.

How does AI speed up audience research?

It sweeps and clusters at a scale people cannot match. We used AI to gather what purpose-led leaders in Sydney were publishing and arguing about, then to group that into themes we could programme against. Work that would take an analyst weeks took days, which is what made a 7-week build realistic.

Can AI keep a brand's voice consistent across a big team?

Yes, if you give it the brand rather than a prompt. We built a custom brand GPT and ran every asset through it, so 14 people producing in parallel still sounded like one brand across web copy, ad creative, partner comms and social. Voice consistency is the first thing that breaks at speed.

What channels work when you cannot rely on paid ads?

Give each channel a job. We ran paid social as a test to identify who responded, then shifted the weight to LinkedIn outreach, partner amplification and direct invitation through existing relationships. Paid told us who to talk to; trusted channels got them in the room.

How small can a team be if it uses AI properly?

Ours was 14 people, most of them students in paid roles through our programme with UNSW and the University of Sydney. AI absorbed the research, list building, verification and drafting. That left the team to make the decisions AI cannot: who to invite, what to teach, what to cut.

How do you make an event keep working after the day?

Treat it as content from the start. We produced the film, the live stream and the social content as part of the system rather than as an afterthought, so the recap, the replay and the talks keep earning attention. The day ends; the content does not.