AI Adoption Isn’t About Models — It’s About Trust (Lessons for AI Builders)

[#1]

If Apple can’t ship AI without backlash, the rest of us need more than bigger models

Here’s what I learned from a decade of building software products at Google, Roblox, and startups:

• AI adoption, like any technology adoption, isn’t about the model. It’s not even about the data.
• It’s about the moment a human decides whether to trust it.

Apple’s recent AI challenges show how hard this really is.

- Users expect perfection

- Any mistake erodes trust and damages the brand

- Even Apple is treading carefully. (link in comments)

Why this matters to Marketplace partners

Startups win on speed, but only if we design for trust.

The real challenge isn’t building the AI. It’s designing the experience around it:

• Where does AI show up? - It has to be integrated into existing workflows to reduce friction.
• What happens when it’s wrong? - We should only apply it to use cases when it's ok to be wrong.
• Does the user understand what’s happening and why? - We need to build interfaces that make it easy to understand the "why" without overwhelming the user.

AI isn’t magic. It’s a product design problem.

Questions for the community

• Which single workflow in Jira/Confluence is least ready for AI, and why?
• What’s the worst AI‑generated failure your users have reported so far?
• How do you build trust with your users, given that AI is not always right?

Drop your lessons (and cautionary tales) below — let’s build a stronger, safer AI layer for every Atlassian team.

Author disclosure
I’m Ala Stolpnik, CEO at Wisary. The perspectives above come from building AI‑powered flows inside Confluence. If you’d like to see one in action, here's our Marketplace listing.

[Ala _Wisary_]

Reference: Apple’s AI strategy is being questioned for exactly this reason: building trust is hard, even for the world’s most loved brand.

Link: https://www.cnn.com/2025/03/27/tech/apple-ai-artificial-intelligence/index.html

[Project Science Ltd]

Good points.  AI is about trust - and trust has to be earned - which is where explainable AI comes in.  We (Project Science Ltd) build AI / ML apps that explain their outputs, such as Predict for Jira.