Access All Areas: Is AI redefining User Interfaces and how we access our data?

What triggered this article?

Whilst working on a personal project recently, I was trying to access all of my social accounts from a central application that I wanted to build. The process was clunky, to say the least. I was universally required to register as a developer and approach integration as though I was building a global marketing tool to resell to multiple users, not for a single user wanting to manage or filter their own history or run personal automations. I have attached an image of the tools LinkedIn provides for API access, perhaps misleadingly called 'Consumer APIs'.

It was this experience that triggered me to write this article. Most platforms still assume a binary world: the UI is for users, and the API is for enterprise developers. But AI changes that.

With LLMs writing code, translating docs, and abstracting complexity, there's no reason a user, even one without coding experience, couldn't say, "Hey, I want to get all my posts from 2022 that mention a certain topic, and create a summary." And yet, most systems simply aren't built for this kind of access and I'm not sure most are ready.

Vibe Coding changes our access methods:

As AI continues to break down the barriers to building software, we're witnessing a shift in how users interact with platforms, systems, and particularly their own data. What used to be the domain of frontend interfaces and point-and-click dashboards is now evolving into a world of 'micro interfaces' and API-first access. Users who are not necessarily developers in the traditional sense, now require more granular, programmable control.

It's time for organisations to rethink who gets access, how they get it, and what they can do once they're in. There are many layers of responsibility required and I think it's going to create many challenges.

From UI to API: An Evolution in User Experience

For decades, user interfaces have been designed around human interaction; buttons, forms, dashboards. They're designed to be intuitive, visual, and often deliberately limited in scope to avoid overwhelming users.

But there is a shift to data led access and we're not just talking about professional developers anymore. Business analysts using AI tools to automate tasks, creators spinning up websites using natural language, or even individuals asking agents to query their own data across multiple systems. According to Anton Osika of Lovable, 2.5 million websites were built on their platform in June 2025 alone, representing 10% of all new sites that month. Many of those users weren't coders. They were everyday people with a goal and an AI assistant.

This access to vibe coding development tools means people are building applications to solve very specific problems; micro interfaces, if you like. And these tools often need data and functionality that platforms have locked behind more traditional, enterprise-style APIs.

The interfaces of tomorrow won't just be human dashboards. They'll be APIs, endpoints, event streams, and developer portals built for a broader, less technical, but no less ambitious crowd.

The Future: Agent-Friendly, API-Native Everything

In a world of personal agents, custom tools, and AI-built apps, users will increasingly want to access their own data in new ways:

• Filtered queries
• Real-time interactions
• Data orchestration across platforms
• Automation without middleware

To support this, we'll need to rethink how we build platforms. Not just by offering APIs as an afterthought but by treating them as a primary interface, with the same care and usability as traditional frontends.

I saw this having an impact in four areas:

1. Security: Trust but Verify, Always

More programmatic access means more attack surfaces. As users delegate tasks to agents, or connect apps via tokenised APIs, platforms will need to get smarter about:

• Fine-grained permission controls
• Time-limited or single-use credentials
• Agent identity and action transparency
• User-friendly revocation and audit tools

The days of "full read/write access" tokens lasting for years should be over. We need secure, context-aware access models that work at machine speed.

We also need to ensure that security controls and safe guards remain in place to assist these users. With API style access comes a different type of risk, it's my view that the users are going to need assistance with security and ensure their accounts remain safe. I have seen many people talk about the tech debt of vibe coders but it is not just the users who will be impacted if bad actors are running amok within our systems and social accounts.

2. Filtered Access: Give Me Just What I Need

Why should an agent have to trawl through 5,000 messages when it only needs the ones about invoice disputes? Platforms need to support more advanced, user-friendly filtering through structured endpoints or even prompt-style inputs.

This opens the door to smart agents that can act like middleware: "Fetch my recent logins, flag any that look dodgy, and send me a daily summary." Not every user wants a firehose. They just want control.

3. Simplified Developer Tools: Docs for the Non-Developer

What if your API docs didn't assume the reader was a backend engineer?

We need to create developer interfaces that cater to the AI-assisted, code-curious user. That means:

• Natural language queries
• Code examples in plain English
• AI-generated walkthroughs
• Sandbox environments built into user dashboards

The next big app might be built by someone using a chatbot and copy-paste. Let's not gatekeep the process with outdated developer assumptions.

4. Infrastructure and Support: Always On, Always Flexible

With the rise of agent-based interactions, your platform isn't just serving humans between 9-5, it's responding to automated calls at all hours. That means infrastructure needs to be:

• Scalable by default
• Resilient to spiky agent traffic
• Monitored for abuse and anomalies
• Backed by support that understands integration, not just UI problems

Support should know what a webhook is. Customer success should know how to onboard an agent. If your platform can't explain how to set up an automated sync between a user's data and their AI, you're behind.