What building Omi is teaching me about designing for uncertainty

Here's a strange fact about first-time home buying in the UK: there has never been more information available about it, and buyers have never seemed more lost.

Every question has a hundred answers. Affordability calculators on every bank's website. Guides from every money site.

Forums, YouTube explainers, government schemes with their own documentation.

A first-time buyer can spend three months reading and finish knowing more facts than their parents ever did, while being no closer to answering the only questions that matter to them personally: where do I actually stand, what's holding me back, and what should I do next?

I'm building Omi, an AI companion for exactly these buyers, and this abundance paradox is the problem underneath the problem.

I'm writing this partly to think out loud, because the honest state of the project is that I'm learning faster than I'm shipping. Some of what I've learned has already changed the product's shape.

The first framing was wrong

Omi started narrower than it is now. The early idea was mortgage readiness: assess a buyer's situation, produce a readiness score, show them how to improve it. Clean, buildable, easy to explain.

It took embarrassingly long to see why it was too small.

Mortgage readiness is one chapter of a journey that includes saving strategy, understanding true affordability, choosing where and what to buy, navigating solicitors and surveys and the grinding months between offer and keys.

A readiness score answers one question and then goes quiet, right when the user's uncertainty is just getting started.

Worse, I began to worry about what a score does psychologically. A number without explanation manufactures certainty it hasn't earned.

Score someone 72 and they'll feel reassured or anxious, and either feeling is unjustified, because the number by itself carries no meaning about their life. What does 72 mean?

What's it made of? What would move it? A score that can't answer those questions isn't clarity.

It's a mood generator wearing the costume of clarity.

So the framing shifted. Mortgage readiness became a feature. The product became something broader: a companion for the whole journey, built around three promises.

Know where you stand. Know what's holding you back. Know what to do next.

Signposts versus a compass

The mental model I keep returning to is the difference between a signpost and a compass.

The internet has given first-time buyers infinite signposts, each pointing somewhere real: here's how deposits work, here's what a broker does, here's ten tips for your credit score.

All true, all generic, all indifferent to whether this particular signpost is the one you need right now.

What a buyer lacks is a compass, something that knows their situation and orients them within it. Not more information. Better orientation.

The distinction sounds subtle and it changes everything about what you build. A content strategy produces articles.

An orientation strategy produces status, prioritisation and next actions, personalised to one person's actual circumstances.

In practice, the MVP I'm working towards looks like this. Onboarding that asks about your real situation, income, savings, timeline, circumstances, not to fill a database but to make everything after it about you.

A view of where you stand across the journey, in plain language. Blockers, named explicitly, because a vague sense that "something's not right yet" is one of the most corrosive feelings in the whole process.

And next best actions, ranked, with reasoning attached.

That last part carries a rule I've adopted: an action has to explain why it matters and what changes when it's done. "Reduce your credit utilisation" is a signpost.

"This is likely your biggest blocker; addressing it improves what lenders see when they assess you, and once it's done, your deposit timeline becomes the thing to focus on" is a compass reading.

The second one takes far more design and product work. It's also the only version that actually reduces uncertainty rather than adding to the pile.

The part where I have to be careful

Omi has an AI layer, and this is where designing for uncertainty gets sharp, because the stakes aren't abstract. People may act on what this product says, with the largest sums of money they've ever handled.

Two boundaries have become non-negotiable. The first is regulatory and ethical: there's a real line in the UK between educational guidance and regulated financial advice, and Omi has to live on the right side of it.

That's not just compliance hygiene. It's product truth.

Omi's job is to help someone understand their situation and get oriented, and to know when the right next action is "speak to a qualified adviser, and here's what to bring to that conversation." Making the handoff to humans a designed feature rather than a failure state is, I think, one of the more important decisions in the product.

The second boundary is about honesty under uncertainty. An AI that answers everything confidently is easy to build and quietly dangerous in this domain.

The harder, correct behaviour is bounded: stay within what the product actually knows, show what an answer is based on, and say "I can't tell from what you've shared" when that's the truth.

I wrote elsewhere that AI products need a trust moment, not just an aha moment. Omi is where I have to live by that with real consequences attached.

And a personal note on credibility, because I'd rather state it than have it discovered: I'm not a mortgage adviser, and I'm not building this from the memory of my own completed purchase.

What I can offer is the discipline of building openly, research with actual first-time buyers, input from people who are qualified in the domain, and a willingness to publish what I got wrong.

The readiness-score story above is the first entry in that ledger. I doubt it will be the last.

What this project keeps teaching me

The transferable lesson, the one that applies well beyond property, is about what clarity means in high-stakes products.

I used to treat clarity as a property of communication: plain words, clean layout, no jargon. Omi has taught me that in a high-stakes journey, clarity is really a property of orientation.

Users aren't confused because your sentences are complicated.

They're confused because they can't locate themselves: they don't know where they are in the process, which of the hundred true facts applies to them, or what to do first.

You can write the plainest copy in the world and leave that confusion fully intact.

Which means the design work is prioritisation, not simplification. Anyone can list twelve things a first-time buyer should think about.

The product earns its existence by saying: for you, right now, it's this one, and here's why.

I don't have the finished version of this. What I have is a clearer definition of the goal than I started with: clarity is helping someone make the next good decision.

Everything in the product either serves that or it's decoration.

If you're a first-time buyer, or you were one recently, I'd genuinely like to hear one thing: the point in your journey where more information stopped helping, and what you actually needed instead. Those moments are the map I'm building against.