Before any question is answered. Applies to every user, every session, authenticated or anonymous, without exception.
The Clarity Engine has a non-negotiable compliance obligation that must be satisfied before the first message is ever exchanged. DISC-01 and DISC-03 require AI identity disclosure and educational purpose disclosure before any mortgage-related guidance is provided. The Colorado AI Act, effective June 30, 2026, requires the same. The PRD is explicit: "AI identity disclosure: visible on all screens."
This requirement applies to anonymous users. Account creation is not a prerequisite. A user who never creates an account has the same disclosure rights as an authenticated user. The system cannot wait for a login to satisfy this requirement.
The disclosure lives in the interface, not the conversation. It renders automatically at session open, before the first message, for every user regardless of authentication status. It is always visible. It never interrupts the conversation. It never requires a click to dismiss. It never repeats inside the chat unless a user directly asks.
Pending Dickinson Wright final sign-off:
"PreFi helps you understand your mortgage options: clearly, privately, and without anyone trying to sell you something. This is an AI tool. Educational only. Not a lender. Not financial advice."
The promise comes first. "Clearly, privately, and without anyone trying to sell you something" is the most important thing a Skeptical Independent needs to hear in the first three seconds. The compliance language follows immediately and completely. This is the first sentence of the PreFi brand. It must read like one.
This is the foundation. Every answer below assumes this disclosure is in place.
"What's on your mind about your home?"
This question works because of what it conceals as much as what it reveals. The user doesn't yet know what this product can do. They think they're starting a conversation. The product already knows its property value, equity, and loan balance. That asymmetry: held quietly until the reveal: is the source of the property reveal's power.
The opening question must never hint at what's coming. No "let's see what we can find." No "we may already have some information about your property." Silence. Then the reveal. The gap between this question and what happens next is where the first moment of trust is built: not through words, but through the shock of competence arriving unexpectedly.
The reverse psychology options: "most people who look into refinancing don't end up doing it": are strong second-beat moves. Earn the relationship first. The product's first job is to listen. Its second job is to astonish.
React only. Never pre-load trust or disclaimer language inside the conversation.
Products that pre-load disclaimers signal insecurity: they argue for their own trustworthiness before anyone questions it. That is the behavior of something that expects to be doubted. And when you expect to be doubted, you invite it. The foundational UI disclosure handles the compliance obligation before the conversation begins. The conversation itself stays completely clean.
The property reveal is the trust mechanism. The user types one address, and the product instantly surfaces their home value, equity, and loan balance. That moment: the product demonstrating it already understands the user's situation before asking for anything: does more for trust than any disclaimer ever written. The user stops evaluating whether to trust the product. They start wondering how it already knows this much. That repositions the entire relationship in the product's favor.
When a user directly asks a trust question: "Are you going to sell my data?": answer immediately, completely, and move on. One sentence. No qualifiers. No elaboration.
"No. I'm an AI. I help you understand your options and tradeoffs so you can decide what's actually possible for your situation."
When someone asks this question, they are testing the product's honesty. A short, direct, unhesitating answer signals the product has nothing to hide and nothing to prove. Restraint here reads as authority. The product should never sound like it's explaining itself. Explaining is defending. Defending implies something to defend against. Answer the question. Stop. Move on. The brevity is the trust signal.
"So you can decide what's actually possible for your situation" is the critical phrase. It frames the product as being in service of the user's decision: warm without being soft. It describes a specific outcome the user can immediately visualize and want, not a general intention. And it makes the user feel individually considered rather than generically answered.
Even though AI identity is already disclosed in the interface, when a user asks this question directly, they are asking for human confirmation in the moment. "I'm an AI" gives them that honesty, and the product never tries to appear to be something it isn't.
"Something changed that may matter for your situation: [specific trigger: rates moved / your property value updated / a program you may qualify for is now available]. Thought you'd want to know. No pressure: everything from last time is still here whenever you're ready."
"Something changed that may matter for your situation" activates curiosity and signals the product is working for the user: without the surveillance connotation of "I've been keeping an eye on you." This distinction matters enormously for the Skeptical Independent and veteran archetypes who are most sensitized to anything that feels like monitoring. The product's watchfulness is demonstrated through the specificity of the trigger, which is not announced.
"Thought you'd want to know" maintains Calm Confidence: no urgency, no pressure, a trusted advisor passing along something relevant. "No pressure: everything from last time is still here" removes all activation energy. The user doesn't have to start over. They just have to look.
Equally important: protect the silence between outreaches. A product that reaches out constantly trains the user to ignore it. A product that reaches out only when something genuinely changed trains the user to pay attention every single time. The restraint between messages is as important as the message itself.
From the product builders who set the bar this product is measured against. This section is not designed for direction. It is the standard against which every design decision gets tested before it ships. Everett owns the how. This section defines the why and the what at the level of felt experience.
The foundational disclosure is not a legal obligation dressed up in brand language. It is the first thing a user reads. It is PreFi's first impression, first handshake, and first promise: all in one moment. A disclosure that reads like a terms of service was designed by someone solving a compliance problem. A disclosure that reads like a PreFi sentence was designed by someone who understands that the first three seconds determine whether everything that follows gets a chance.
The revised language: "PreFi helps you understand your mortgage options: clearly, privately, and without anyone trying to sell you something": leads with the promise, not the constraint. The compliance language follows immediately. Both obligations are satisfied. One version makes the user lean forward. The other makes them scroll past. Design the one that makes them lean forward.
The reveal is the first proof that PreFi is unlike anything the user has ever encountered in the mortgage space. It needs to be directed, not just implemented. There is a profound difference between a feature that works and a moment that is felt. This needs to be felt.
The sequence: the user enters their address. A deliberate pause of approximately 800 milliseconds follows: not a spinner, not a progress bar, a considered beat that communicates the system is doing something real. Then the property appears in sequence. The address first. The aerial image second. Then: with a slight delay that makes the numbers feel discovered rather than populated: the value, the equity, the loan balance. Each arrives as if being read to the user rather than displayed.
The AI response that accompanies the reveal is the only moment in the entire product when it is allowed to be quietly proud of itself. The anchor arrives at exactly the right moment: not rushed, not delayed. As the last number appears, one beat, then:
"Here's what I found."
Then silence. Let the numbers speak. The next prompt follows quickly enough that the user doesn't wonder what to do next: but not so fast it feels automated. That timing is a build specification, not a developer judgment call.
The Magical Moment fires when the Advisor detects sustained confusion across two or more consecutive exchanges. It is not a stage. It is a conditional intervention. Not every session triggers it: and that is correct. When it does fire, three things happen in sequence: a 500ms fade transition creates visual breathing room, the most complex visual on screen is replaced with its simplest equivalent, and an archetype-specific analogy card appears.
"$450 a month is a car payment you don't have to make."
The moment should feel like a breath, not a celebration. No confetti. No congratulations. A quiet clearing followed by one plain-language sentence that makes everything click.
This moment requires a complete spec from Everett before BIX builds anything: trigger conditions, animation timing, card design, analogy library by archetype, and reduced-motion fallback. This is a Discovery deliverable. It is a Go/No-Go criterion for Alpha. If it isn't designed, it will be approximated, and approximated is not acceptable for the most important moment in the product.
Some users will question the product before they engage: asking "what is this?" or "why do you have my information?" before entering their address. The current architecture handles users who question after engaging. This handles users who have questioned before.
A single grounding sentence activates only when early skeptic signals are detected. Never offered proactively: offered proactively it reads as defensive. Offered in response to a signal it reads as honest.
"I help you understand your mortgage options without talking to a lender."
One sentence. No disclaimer energy. Answers the concern and immediately reframes the product as the antidote to what the user is worried about. "Without talking to a lender" does the most work: it names the thing the Skeptical Independent fears most and removes it in four words.
The product is not what it does.
The product is how it feels to use it for the first time.
Every design decision at Everett gets tested against that sentence before it ships. If a decision makes the product more capable but less appealing, reconsider it. If it makes the product feel inevitable: like it could not have been designed any other way: it's right.
| Questions 1, 2, 4 | Ready to lock and incorporate into Figma spec. |
| Question 3 | Ready pending Dickinson Wright sign-off. |
| Section 5: Product Standard | Active for Everett Week 2 design work. |
| PEWC confirmation | Required with BIX before the Always Approved feature ships. |
| Magical Moment spec | Required from Everett before BIX Sprint 1. |