The three jobs AI is doing well
1. Strategy generation
Writing a real campaign strategy — property positioning, target buyers, creative direction, full ad copy — used to be the slowest part of running a serious campaign. AI now drafts the heavy lifting; strategists refine it. The throughput unlock is the whole point. Inside Agentis this is how every strategy doc is built.
2. Live audience optimisation
Modern campaigns reshape themselves in flight by what each audience actually does — what they watch, what they click, what they save. The agent doesn't have to read the data and make the call. The system makes it; the agent reviews it.
3. Cross-tenant signal
This is the hardest part to build and the most defensible once it exists. Anonymised audience-response signals from every campaign feed into a shared model — the Hive Network — that makes targeting on the next campaign sharper than it would be from any single account's data. It compounds.
What AI isn't doing well (yet)
- Negotiation. AI can summarise homeowner sentiment and draft pricing scenarios. It cannot replace the agent at the auction floor or on the phone late at night.
- Creative judgement. AI generates variants. A human still picks which photo of the kitchen actually sells the lifestyle.
- The listing pitch. No homeowner signs based on a slide deck or a chatbot. The pitch is still a human-to-human conversation about trust.
What this means for NZ agents in the next 12 months
Agents who adopt these tools now will compound faster than those who wait. Not because AI alone wins listings — it doesn't — but because the time saved on strategy + audience analysis can be reinvested into the parts of the job that do win listings: relationships, presentation, negotiation. The 2026 guide covers the practical adoption path.
If you want to see what an AI-augmented Agentis campaign looks like for a property in your patch, run a free Suburb Audit or book a strategy preview.