AI INTAKE / HELP CUSTOMERS FINISH

Help customers finish.
Know what needs fixing.

An AI intake had promise, but customers could lose their place or get stuck. I turned the problems into a clear repair plan the team could act on.

Read the case study

PRODUCT STORY / AI INTAKE EXPERIENCE

A high-trust journey needs more than a smart conversation.

The experience connects conversation, captured answers, review, and document completion into one customer journey. The case study shows how I shaped the strategy and readiness model around the moments where trust is won or lost.

A working AI feature
is not the same as a trusted product.

I helped a confidential financial-services client turn a promising AI intake experience into a disciplined product recovery plan: preserve what worked, identify where trust broke down, and define the evidence needed to move forward.

Discuss a readiness review

WHAT I BUILT: AI Product Recovery & Readiness Architect
THE CONTRIBUTION: Strategy, UX research, technical reasoning, evidence design & remediation criteria
THE VALUE: A clearer path from inherited AI product to trustworthy customer experience

Public portfolio version. Client and product identity are anonymized, and sensitive implementation details are abstracted.

01 / The mandate

Turn uncertainty into a product decision.

A confidential financial-services team had a working AI intake experience with real momentum: conversation, structured data capture, document generation, signing, and a clear ambition to modernize a high-trust customer journey. What they needed next was not another generic AI opinion. They needed a way to know what was valuable, what was fragile, and what had to be true before the product could move forward with confidence.

I came in as the product recovery lead across strategy, UX research, technical assurance, and acceptance design. The goal was to protect the strongest parts of the investment while giving stakeholders a clear, evidence-backed path for the next phase.

Consultation workspace with an advisor, transcript, saved answers, and section review — A public-safe reconstruction of the core intake relationship: conversation, captured answers, and advisor review.

02 / The customer reality

Trust breaks when the user loses control.

The product was asking people to complete a sensitive financial journey through an AI-guided experience. That meant the interaction had to do more than sound natural. It had to remember answers, recover from friction, expose the right controls, and keep the customer oriented from the first question to the final document.

The most important issues were not isolated screen defects. They were confidence defects. If a customer cannot see what has been captured, correct a field, or continue when the conversation stalls, the experience stops feeling intelligent and starts feeling risky.

Comparison of a conversation-only mobile state and a recovery pattern with answer review — The case for recovery: an AI journey needs a visible way to inspect, correct, and continue.

03 / The insight

The product lived between AI, UX, and evidence.

The breakthrough was treating readiness as a connected system, not a single test result. Conversational behavior, saved state, responsive layout, generated outputs, signing flow, and support controls all had to work together. A problem visible to a user could originate in interaction design, state handling, orchestration, or an unclear acceptance standard.

I connected multiple evidence streams into one decision model: product walkthroughs, source-informed technical review, user testing, cross-device experience checks, and output evaluation. That gave the client a way to separate noise from signal and focus the team on the changes that would actually increase confidence.

Evidence lanes connecting implementation, user experience, output quality, and readiness decisions — A readiness model built for AI products, where experience quality and system behavior have to be judged together.

04 / The strategy

Preserve what worked. Redesign the trust layer.

The recommendation was not to throw everything away. The foundation had product value. The opportunity was to make the experience more inspectable, recoverable, and measurable so the client could keep moving without mistaking a promising prototype for a dependable journey.

I shaped the work around three product principles: one shared intake record, visible recovery at the moment of friction, and acceptance criteria based on complete journey behavior. That turned scattered findings into a practical roadmap for product, design, and engineering decisions.

01One intake state
Conversation and manual edits resolve into one reliable source of truth.
02Visible recovery
Customers can inspect, correct, retry, and continue when the experience needs help.
03Journey evidence
Progress is judged by the complete customer path, not a single successful interaction.

Sequence showing answer capture, saved state, confirmation, and next-question logic — The core principle: capture the answer, confirm the state, then advance the experience.

05 / Interaction architecture

Make the AI experience editable, visible, and resilient.

A high-trust intake cannot depend on a conversation alone. Customers need to know what the system heard, what it saved, and what still needs attention. That meant designing requirements for review, correction, retry, and continuation across screen sizes.

The mobile experience made the principle especially clear. The fallback could not live somewhere else. It had to be available where the customer was working, so a person could recover without losing momentum or confidence.

Mobile states showing conversation, answer review, and field correction — A mobile-first recovery pattern that keeps the customer in control of the intake.

Customer using a landscape tablet with advisor and form visible side by side — A larger-screen view of the same principle: keep the conversation and captured information close enough to preserve confidence.

06 / Delivery system

Convert critique into work a team can execute.

The value of the engagement was not just finding issues. It was translating them into a shared working system: what the customer should experience, what evidence supported the finding, what the correction needed to prove, and how the team could verify completion.

That handoff covered conversational continuity, cross-device completion, manual corrections, output quality, signing evidence, service controls, monitoring, and readiness testing. It gave stakeholders executive clarity and gave the development team specific outcomes to build toward.

Remediation requirement with evidence, expected behavior, acceptance criteria, and validation steps — A repeatable requirement format that moves AI product feedback from opinion to accountable execution.

Financial form structure populated with sample values and a signature field — The outcome layer matters too: the completed artifact has to reflect the journey that produced it.

07 / What changed

The client gained a path, not just a report.

The engagement reframed the product from a question mark into an actionable recovery plan. Stakeholders could see what was worth preserving, where trust was breaking down, and what evidence would determine the next decision.

The business gained a way to decide what to repair first and what to verify before asking customers to rely on it. That makes the next investment clearer and gives the team a shared definition of a journey customers can finish.

Readiness process from assessment through remediation and validation — The outcome: a clearer route from inherited AI experience to evidence-backed product confidence.

AI PRODUCT RECOVERY / TRUST ARCHITECTURE

Bring clarity to the AI product you already have.

If an inherited AI experience has promise but needs sharper direction, the work is not just critique. It is strategy, evidence, product judgment, and a clear definition of what trust should look like in the finished journey.

Discuss a readiness review Explore the audit approach