Work
with Else

Want to further a new product idea, tackle a difficult business challenge, or explore an opportunity? We’d love to help.

hello@elselondon.com

Explore a
career with Else

Ready to put in everything you have to make your mark? We promise to make your time at Else the highlight of your working life.

talent@elselondon.com
Insights

From AI-retrofit to AI-native

Takeaways from SXSW 2025

This year’s SXSW made one thing clear: AI is no longer an optional add-on. It’s infrastructural.

Over five intense days in Austin, across thirty sessions spanning design, strategy, tech and ethics, the mood had shifted. Just after OpenAI launched ChatGPT in Nov ‘22, SXSW 2023 was about hype. 2024 was about getting ready. But 2025? Well, we’d say that this was the year reality bit.

The thing that stuck was the idea that “this is the era of AI-native, not AI-retrofit.” And once you’ve seen the difference, you can’t unsee it.

Many companies are still playing with AI at the edges, bolting on AI tools to existing workflows, riding AI side-saddle, commissioning CoPilots.

But the others? Well, it seems, they’re rebuilding business architecture entirely.

As Neil Redding explained in ‘The Auto-evolving Business: AI’s Agentic Near Future’, we’ve already moved from Prompt to Participate, to Delegate, and, for some, to Initiate—where AI doesn’t just suggest; it initiates and orchestrates. IBM is doing this across its HR function, while Adobe is doing it in real-time design systems.

Okay, these are huge tech companies, but you can see it coming.

So AI is not a sidebar. It’s not an R&D effort. It’s becoming the new plumbing of business.

Photo by MJ Tangonan on Unsplash

Three takeaways from SXSW 2025

1. AI is no longer optional or adjacent — it’s infrastructural

Taking Redding’s framing, businesses are moving quickly through the Prompt > Participate > Delegate > Initiate > Symbiosis pathway, and we’re beginning to see businesses where AI is operating at the organisational core — not just tools augmenting existing working practices, but reshaping how work gets done and services are structured, orchestrated, and delivered. This demands new design patterns, new trust models, and new organisational muscles.

SXSW would have you believe that this is happening, everywhere and now! Now! Now! But while we can always temper the collective enthusiasm with the reality that this will take ‘regular’ businesses more time, it was clear that early movers have moved. We know that hundreds of roles were replaced by operational AI at Klarna, we know IBM’s HR system has gone agentic, and we also heard that Deloitte has launched 150 agents to ‘support’ 80K employees.

As Josh Clark said in Sentient Design: AI & Radically Adaptive Interfaces, we’re now designing relationships between people, systems, and autonomous intelligence. It’s no longer UX, it’s shifting to experience infrastructure.

“We’re designing environments, not interfaces. Context-aware, emotion-sensitive, multimodal systems that collaborate — not just respond.”

2. This era is AI-native, not AI-retrofit

Microsoft’s Matthew Duncan, from their Future of Work team, shared his research and unpacked what makes an AI-native company: flatter teams, faster feedback loops, and systems built to learn, not just operate. These firms are not bolting AI on — they’re building with it from first principles.

He talked about a company called Moon Base Croissant — an AI-powered boulangerie in Paris using AI to optimise baking, logistics, marketing, and service design. It’s a croissant company moving like a tech startup, succeeding in a culturally entrenched market and looking to open two more locations soon, totalling four.

As he put it:

“AI-native startups are hitting revenue milestones faster than any SaaS company before them.”

This distinction between AI-Native behaviour and AI-Retrofit struck a chord, especially when he referenced the behaviour of Barnes & Noble in the mid-nineties. B&N saw the emerging internet as a channel adjacent to their business, whereas Amazon thought internet-native and saw a new emerging platform.

Which leads me to wonder, who right now is having a Barnes & Nobel moment, retrofitting AI like they did with e-commerce — or rethinking their game like Amazon?

 

3. The next era of design isn’t human- or AI-centred — it’s context-centred, trust-sensitive, and participatory

This was the most complex theme to emerge, and concerns the ‘what’ of it all. What does a post-AI adjustment look like for a business?

Meredith Whittaker (Signal) delivered a line that resonated in the room:

“The only way to protect data is not to collect it.”

Which sets the context for our AI interactions and points to the trust paradox of agentic systems. For in order to act on our behalf successfully, we’re going to need to give these tools large amounts of personal data (so they can get it right). Which in turn necessitates new levels of trust designed into those interactions, products and services.

We hear a lot about agentic systems on the enterprise side, but we should also remember that we’re also at the advent of Personal AI, where individuals have their own agents interacting with enterprise agents.

But most organisations haven’t earned it. Cory Doctorow called out the mismatch between what AI tools demand (permission, agency, autonomy) and what they give back (value, accountability, clarity).

Product/service innovators have a huge role here. Not in checkbox compliance — but in creating dynamic, trust-aware consent systems, agentic interactions, and inclusive models that don’t just reflect the world — they improve it.

 

“We’re not just designing interactions — we’re designing intelligent environments, trust architectures, and shared agency between people and machines. The canvas has changed, and so must we.”

So, where does that leave us?

At Else, we’ve been designing AI-centric services since 2018 — across trust-led identity platforms, intelligent CX, and ethical product experiences. But the scale, pace, and stakes have changed.

These takeouts aren’t just directional. They’re transformational.

If you’re not already asking:

  • How does AI reshape our services?
  • What would an AI-native version of our business look like?
  • And how do we build trust, not just tech?

…then now’s the time to start..

We think it’s time to stop waiting for playbooks — and start prototyping new ones.

Getting started
Find out more about how Else can help.

Let's get started