About This Event

Agentic Commerce, Melbourne was held on Thursday 1 October 2026, 5:30pm to 8:30pm, at the Mantel Hub, Level 2, 452 Flinders St, Melbourne, hosted by Mantel Group with Prosperity Media supporting the night. It was the Melbourne leg of Kindred Talent's national Agentic Commerce Series and the first time the series had left Sydney: Brisbane followed on 8 October at The Precinct and Sydney on 22 October at Iru, Salesforce Tower. Each city got a locally relevant panel mixing SEO, paid media and engineering. Sixty seats, free for industry peers, aimed at founders, C-suite and heads of department across eCommerce, digital, SEO, performance, CX, data and engineering.

The evening ran from doors at 5:30pm through an Acknowledgement of Country and introduction at 6:00pm, panel discussion from 6:05pm, audience Q&A at 7:15pm (questions submitted at registration were worked into the night), and networking and drinks from 7:30pm.

Topic Focus

The series premise: when an agent is the decision-maker, the rules change. AI agents are moving past assisting decisions toward making purchase recommendations, and sometimes the purchase itself, on behalf of consumers. The staging was deliberate: the marketers fighting for visibility inside AI answers on one side, the engineers building the systems that do the choosing on the other, and the disagreement as the point.

The shifts under discussion: SEO matters little if an AI never crawls the page; ad creative matters little if there is no human to see it; feed quality, structured data, reviews and brand authority become the new battleground; and attribution breaks when there is no trackable click.

The Discussion

Three themes, framed by the organisers as not what is coming but what to do about it on Monday:

Getting found: visibility when the buyer is a machine. What signals decide whether an assistant recommends you over a competitor, how much of the old playbook still applies, and what you can actually influence today.
Getting through: from recommendation to conversion. What happens to your funnel when the thing reading your product page will not be moved by a hero image, the comparison happens somewhere you cannot see, and the click you used to measure never arrives.
Getting it built: where marketing meets engineering. What marketers need from engineering, what engineers need from marketing, and how the two ship something into production together instead of trading tickets across a backlog.

Getting found opened by settling what agentic commerce actually means and where an agent's trust hits limits, put to Lovee as the one panellist who builds agentic systems for a living. It moved through where agentic commerce starts for an SEO team, whether a site's CMS changes how LLMs detect a brand, and how REA decides what crawls what: realestate.com.au's robots.txt blocks GPTBot, ClaudeBot, PerplexityBot, CCBot and other AI bots while leaving OpenAI's search and user bots free to crawl. Rahul was asked how much of the old playbook still applies now that his title says GEO. The theme closed on ads starting to appear inside ChatGPT, and whether paying to be in the answer changes AEO and GEO.

Getting through went deepest on REA. Realestate.com.au has gone from filters on the website, to natural language search after a 12-month trial, to conversational search, to an AI assistant, to putting its listings inside ChatGPT, which Mediaweek called Australia's first ChatGPT property search app. The panel covered how guardrails were set up inside the business, then turned to retail: what customers actually ask AI shopping assistants, where the sale dies, and how technical the fix is, from structured data to Merchant Centre.

Getting it built asked which businesses are ready for agentic commerce without knowing it, then worked through the constraints. Australia ruled out a text and data mining exception in October 2025, so there is no copyright carve-out for training AI on your content, and from 10 December businesses must explain in their privacy policy when automated decisions significantly affect people. It covered who should own the work across marketing, product and engineering, where to start, how to attribute commercial wins, and how far agents can be trusted after an OpenAI agent recently went rogue. Each panellist closed with one thing to do on Monday.

The Panel

Four panellists, one moderator:

Lovee Jain, Agentic Systems Engineer, IAG
Nitin Karunakaran, National Head of SEO, Alley Group (part of Nunn Media)
Rahul Sengupta, Melbourne SEO and GEO Lead, Prosperity Media
Jine Wu, SEO Operations Manager, REA Group

Moderated by Wylan Ho, Founder, Kindred Talent.

Speaker

Rahul Sengupta is an SEO and AI-search consultant based in Melbourne, Australia. He has 8+ years in enterprise technical SEO and 12+ years in digital marketing across banking, retail, B2B and eCommerce. He is Melbourne SEO and GEO Lead at Prosperity Media and founder of Rahul Sengupta Digital.

Takeaways

Rahul's thread through the night, drawn from his prep and panel notes:

The old playbook re-ranks rather than dies. Technical foundations, content clarity and structured data still gate you in, but authority matters less than expected: across the 122,787 AI citations Prosperity analysed for 12 brands, the most-cited listicles sit on domains with a median DR of 43 versus 53 for the rest, and 32% of citations went to domains under DR 40.
LLMs cite fresh, localised, plain-English roundups. 90% of tested prompts cited at least one listicle or best-of roundup. The most-cited pages have a median age of 97 days, "Australia" in the title on 55% of winning listicles versus 35% for the rest, and plain English (Flesch 60+) 12 points more common among winners.
Schema is table stakes, not a differentiator. 90% of cited pages carry schema and only about half have a clean heading hierarchy, with no lift for top performers. Freshness, local signals and readability separate the cited from the ignored.
Measurement moves from traffic to influence. AI citation share of voice as the leading indicator, branded search lift as the bridge, assisted organic conversions as the north star, the same stack he prepared for the SEOFOMO panel in March, now with a citation dataset behind it.
The Monday-morning action. Run a prompt-set audit of what LLMs actually cite in your category, fix freshness and localisation on the pages that should be winning, and wire citation share of voice into stakeholder reporting.

Agentic Commerce, Melbourne