TL;DR

  • AI is becoming the front door to software: instead of clicking, you ask.
  • Agents (like those built on Model Context Protocol) go further, skipping native UIs to act on your behalf.
  • But not everything should be replaced: old UIs remain best for precision, safety, and learning.
  • For Kiwi companies (Trade Me, Xero, One NZ), the challenge is balancing AI’s magic with trust, transparency, and compliance.
  • In AWS this was achievable in last decade but now it’s mainstream

1. Language as the new input

Once, we memorised commands. Then we clicked icons. Now, increasingly, we just talk.

Satya Nadella called human language “the new UI” back in 2016. The phrase “AI is the new UI” has since been echoed by business leaders like Terry Jones.

Everyday examples show the shift:

  • Microsoft Copilot lets you type “Summarise this email thread and suggest a reply” without navigating menus (docs).
  • Spotify DJ curates music by conversation instead of endless scroll (launch).
  • Locally, Xero could easily imagine “File GST for last quarter” as a natural command, or Trade Me: “List this couch for $100, write a better description, and suggest delivery options.”

This doesn’t kill UI — it just re-centres it around intent. Buttons still matter for detailed control, but the front door is increasingly conversational.

2. Agents as the new UI

Language gets you partway. Agents take you further.

Instead of just answering, an agent acts. They orchestrate tools, call APIs, and return results — all without you touching a screen. Salesforce calls this the rise of “agents as the interface” (Salesforce, 2025).

The underlying tech: protocols like MCP. These standardise how AI systems access calendars, files, or customer records in a safe, permissioned way (modelcontextprotocol.io). It’s like giving your assistant a swipe card: they can only open the doors you authorise.

For Kiwi businesses:

  • One NZ could let a customer simply say: “Upgrade me to the latest iPhone plan and transfer my number.” The agent checks stock, runs the order, updates billing.
  • Trade Me could let sellers ask: “Re-list unsold items with a 10% discount.” The agent edits and posts, logging every change.
  • Xero could expose invoice tools to an AI layer — your request flows straight into the accounting engine, with audit trails for compliance.

Lookback 🥝

Back in 2020, I built an Alexa skill demo for customer care. At the time it felt futuristic: voice-driven support that cut through clunky menus.

What’s striking now is how close this is to reality for enterprise apps. Thanks to conversational AI plus protocols like MCP, what was a demo hack on Alexa then is turning into a practical product pattern today. The dream of “skip the app, just ask” is no longer speculative — it’s deployable.

Watch the demo

Why this matters

The hype is real, but the nuance matters.

  • AI removes drudgery: fewer clicks, fewer swivel-chair moments.
  • Old UIs still anchor precision: pixel edits, financial approvals, compliance workflows.
  • For NZ firms, the opportunity is to design hybrid systems: conversational entry, agentic execution, and strong visible controls.

The shift won’t erase apps like Xero or Trade Me — it will just make their APIs and policies the true product surface. The UI you see may shrink, but the AI in between will expand.

Written for KiwiGPT.co.nz — Generated, Published and Tinkered with AI by a Kiwi