Building in plain sight: why we can build AI faster than you'd expect

Welcome to the second in our series on how we're building SmartPlans - and why our starting position matters more than most people realise.

‍

People sometimes ask us how a company our size is building AI products this quickly. SmartPlans went from concept to alpha in months, not years. We're iterating weekly with real agencies using it in real assessments. And we're already planning what comes after SmartPlans.

‍

The honest answer isn't that we're faster or smarter than anyone else. It's that we started from a different place - and that place was a decision made nearly ten years ago.

‍

The thing many AI projects spend their time on

‍

Here's what building AI in care tech often looks like. You have an idea - say, generating care plan suggestions from a conversation. Before you can write a single line of AI code, you need data. And for most care technology platforms, the data underneath isn't ready.

‍

Some were built as separate modules stitched together over time - a rostering tool here, a care planning tool there, a finance layer bolted on later. Others started as paper-replacement systems and added features as the market demanded them. The result is the same: data that lives in silos, follows different structures depending on the module, and wasn't designed to be consumed by AI.

‍

So before building anything intelligent on top, you're spending weeks - sometimes months - just cleaning, mapping and standardising the data. That's a natural consequence of how a lot of care tech platforms grew up.

‍

A bet we placed nearly ten years ago

‍

When Birdie was founded, AI in homecare was a distant dream. But the founding team made a deliberate architectural decision: build one platform with a unified, structured data model. Care, scheduling, finance, assessments - all on the same foundation. Every agency on Birdie shares that structure.

‍

At the time, this was about giving care providers a single, coherent system instead of a patchwork of tools. It was about making data usable for reporting, for compliance, for running a better business.

‍

But those same data foundations that make a platform work well for humans also make it ready for AI.

‍

That means when we build an AI feature today, the data is already there, already clean, already consistent. We don't spend months preparing. We go straight to the thing that matters: does this AI feature actually help a care professional do their job better?

‍

It also means we build once and it works for all our care assessment users. We don't have to rebuild or reconfigure SmartPlans for each agency because the assessment structure is shared and so the data model is shared.

‍

We didn't know SmartPlans was coming when we laid those foundations. But we knew that structured, high-quality data would be the thing that let us move fast when the AI moment arrived. That moment is now.

‍

Why this compounds

‍

There's a second advantage that becomes clearer over time. Because all our AI features draw from the same structured data layer, work done for one product benefits the next.

‍

A data quality check built for one feature can be reused by another. A transcript processing improvement for SmartPlans feeds into whatever we build next. Each new AI product starts further along than the last one did, because the foundations keep getting stronger.

‍

Compare that to a platform where every new AI feature has to solve the data access problem from scratch - different integrations, different cleaning pipelines, different validation rules. That's the reality for a lot care tech platforms who have to be retro-fitted for a new AI era.

‍

The bolt-on problem

‍

There's another pattern we see in the market: standalone AI tools that sit alongside a care platform rather than inside it. A transcription tool that gives you a summary, but leaves you to copy the relevant bits into your care management system. An AI notetaker that produces good output but has no idea what your assessment fields are called or how your templates are structured.

‍

These tools can be useful. But they hit a ceiling quickly, because they're working with blocks of text on one side and a care platform, they can't see into one another. Every time the user has to bridge that gap manually - finding the right section, deciding what goes where, copying it across - the AI hasn't finished the job. It's just moved the starting line.

‍

Because SmartPlans lives inside Birdie, it doesn't have that gap. The AI reads from and writes to the same structured data that the rest of the platform uses. It's the reason SmartPlans can suggest content for assessment fields directly, rather than handing you a document and wishing you luck.

‍

And because it's connected to the rest of Birdie, what the AI learns during an assessment doesn't have to stop at the care plan. Need a follow-up action created? A message sent to the care team? A flag raised for review? Those are all possible when the AI lives inside the platform that runs the operation - not when it's sitting in a separate tool.

‍

We're not immune to data quality challenges. But we start from a fundamentally different position, and that advantage widens with every month of structured data we collect across our customer base.

‍

What this means for SmartPlans

‍

When a care professional records an assessment conversation and SmartPlans suggests the answers, every part of that process is faster and more accurate because the data underneath is structured. The AI knows exactly which assessment fields to fill. It knows the dropdown values. It knows how each agency's templates are configured. It doesn't have to guess, and it doesn't have to be taught each time.

‍

That's not a feature we built last year. It's a consequence of decisions made when Birdie was first designed - decisions that are now paying off in ways we couldn't have fully predicted, but were always building toward.

‍

Why we're telling you this

‍

This series is about building in plain sight - showing you not just what we ship, but why we can ship it. The structured data advantage isn't glamorous. It doesn't make for a good demo. But it's the reason SmartPlans exists at the pace it does, and it's the reason we're confident about what comes next.

‍

The best AI products aren't built by the teams with the best algorithms. They're built by the teams with the best data - and a head start on getting it right. Nearly ten years of structured, high-quality care data, collected consistently across every customer. That's not something you can replicate quickly, however good your AI team is. We intend to keep proving it.

‍

— The Kites, Birdie's SmartPlans squad

‍

To learn more about Birdie's AI developments, head over to the Smarter Care Lab for more behind-the-scenes info.

‍