New AI Design Flows: Selling with Real-Time Prototyping and Webflow + AI

The Origin of UX Pilot: From Internal Tool to Product

Not a Planned Pivot Adam is clear that UX Pilot was not born from a deliberate decision to shut down the agency and build a product company. It grew organically from a practical question: now that ChatGPT exists, where can AI be applied to improve the agency's recurring design workflows? The team held internal workshops and brainstorming sessions to explore options.

Notably, Adam's initial position was the opposite of where the product ended up. In early meetings, he explicitly stated that UX should not be used to generate creative assets. AI should help curate, not create. This turned out to be a conviction he would completely reverse within months.

The Wireframe Question The inflection point came during a session with a client. The team had gathered research data from user interviews but needed a fast way to visualise the ideas that data suggested. A client asked whether UX Pilot could generate wireframes. Adam went looking at the market and found no flexible, truly AI-powered wireframe generator. Every existing tool was essentially just a template swapper changing copy or layout slightly. That gap became the product.

Early Goals Were Modest The original commercial target was not to build a venture-backed unicorn. It was to replace one agency project with product revenue. At $10,000 per month per project, reaching that number from UX Pilot by the end of the first year was the goal. The actual outcome far exceeded this.

Growth and Traction

The Moment Growth Kicked In UX Pilot shipped iteratively for the first three to four months without seeing strong adoption. Users were experimenting with features but not forming habits around them. Then a specific update was pushed — Adam does not name the exact change but contextualises it as the combination of two things: a smooth workflow that brought generated designs back into Figma, and consistency controls that let users maintain a coherent visual style across multiple generated screens. After that update, the revenue trajectory went from roughly $3,000-$4,000 per month to nearly $10,000 in a single month, then $15,000, $30,000, $45,000, and continued compounding from there.

Scale as of the Session By the time of this session (April 2026), UX Pilot has:

• 600,000 total users
• 15,000 paying subscribers
• A 30-person team - $5.3 million ARR
• Bootstrapped, no external funding
• A newsletter with 600,000-700,000 subscribers and approximately 35% weekly open rates The SEO Playbook The same content-and-SEO approach Adam used to grow the agency was applied to UX Pilot.

Being early to target keywords like "UI generation," "AI design," and related UX terms resulted in strong search rankings that persist. LinkedIn content was also a major driver.

Contrary to conventional advice not to promote your own tool too heavily, Adam found that sharing product updates — not promotional posts but genuine progress updates — generated high engagement because the audience was already interested in both the AI design space and the specific product's evolution. That engaged community is now reflected in the newsletter open rates.

The Sales Process: Selling with Real-Time Prototyping

The Core Problem: Proving You Are the Right Fit Adam frames the early sales challenge at any agency as a trust problem, not a capability problem. Telling a prospect you have 10 years of experience and showing them past case studies is no longer sufficient. Prospects are harder to convince. The goal of the sales call is not to solve the client's problem in the meeting; it is to demonstrate that you understand the problem and are equipped to solve it.

Wireframes as a Sales Tool Adam's agency began incorporating AI-generated wireframes directly into proposals. After a discovery call, the team would take their notes, run them through UX Pilot, and produce visual representations of what the proposed solution might look like. These were not

polished designs; they were signals of comprehension and execution ability. A simple example: if a prospect wants a new landing page and the designer can immediately flag that the hero section has problems and show a quick visual of what an improved version could look like, that closes the gap between "we heard you" and "we can actually do this."

The Use of Call Transcripts A workflow common among UX Pilot users is to record discovery calls, transcribe them, paste the transcript into UX Pilot, and use that as the prompt context for initial design generation. The tool ingests the unstructured conversation and uses it as the brief, generating first-pass screens that reflect the client's actual language and stated requirements. This bypasses the step of writing a formal brief and speeds up the path to something visual.

Closing Faster, Not Just Iterating More When asked whether real-time prototyping means faster closes or more discovery iterations, Adam is direct: it is primarily about closing faster. The visual shorthand of a working prototype or a set of wireframes cuts through the ambiguity of text-based proposals. Clients can react to something they can see and click rather than something they have to imagine.

Telling a prospect you have 10 years of experience and showing them past case studies is no longer sufficient.

Live Demo Walkthrough

Adam shares his screen and walks through several capabilities of UX Pilot. The following captures what was shown and described.

The Canvas Interface UX Pilot operates on a web-based canvas. Users can prompt from within a "studio" interface, which gives more control over parameters, or they can use a chat interface for a more conversational experience. The canvas supports multiplayer, so clients or collaborators can be invited to view and comment in real time during workshops or review sessions.

Uploading Context and Generating Screens Adam demonstrates uploading a file (representing a call transcript or brief) and prompting the tool to generate an initial screen based on that context. The generation is fast and produces a visual design rather than a wireframe, reflecting how the product has matured from its wireframing origins into full UI generation.

Themes and Design System Controls Users can define a theme within UX Pilot by specifying colours, accent colours, fonts, and border radius values. This maps loosely to a brand identity without requiring a full design system to be loaded. UX Pilot also ships with a set of predefined themes. When a theme is applied, generated designs respect those parameters and produce on-brand outputs rather than generic AI-default styling.

For Figma users, the tool offers a deeper integration: users can sync their Figma components, and UX Pilot will use those components as building blocks inside generated designs. This is a more complex integration suited to teams working on established products with mature design systems.

Prototypes as React Apps Beyond single screens, UX Pilot can generate a full clickable prototype in the form of a React application. Adam demonstrates this: the result is a navigable, multi-screen prototype that works like a real website. Critically, the source code is accessible. Developers can take the generated React project, connect it to a backend, extend it, or port it to another framework. The CSS styling, component structure, and layout are all already done — there is no translation step from a Figma file into code.

The prototype can also be published to a live URL, making it function as an actual web page or web app, not just a design artefact.

Webflow and Figma Export From the UX Pilot canvas, generated designs can be copied and pasted directly into Webflow. Adam demonstrates this live: the design appears inside Webflow where it can be edited, extended with additional pages, and published. He notes a current limitation: if Webflow already has predefined component styles or a design system in place, those may not be captured automatically — local styles are created instead, and some manual linking may be needed.

The same copy-paste workflow applies to Figma. Designs land in Figma as structured, componentised elements with layouts intact, not as flattened images.

Predictive Heat Maps and Accessibility Tools Adam briefly demonstrates a heat map feature that simulates where users would look when viewing a design. In one example, a face in a hero image is drawing significant attention away from the conversion-focused area of the page. Replacing or repositioning that element and rerunning the heat map shows a measurable improvement in attention distribution toward the call-to-action. This type of tool allows designers to make evidence-based arguments about layout decisions without waiting for actual user testing.

The tool also supports accessibility reviews, user flow creation, and diagram generation, positioning it as a broader professional design environment rather than a pure code or visual generator.

Speed as a Design-Specific Requirement Adam makes a distinction between how speed matters differently in design versus development. A developer can trigger a ten-minute build and step away. A designer needs to see results and react immediately; the iteration loop must be tight. UX Pilot offers two generation modes: one optimised for maximum speed, and one that trades a little time for higher quality output but still targets a maximum of 90 seconds per generation. Multiple generations can also be launched simultaneously on the canvas.

Sharing Outputs with Clients

Hiding the Prompts When sharing work with clients or prospects, UX Pilot allows users to share a view that shows the outputs but hides the prompts used to generate them. This is relevant for agencies that want to present work professionally without revealing the AI-assisted process, or who simply want to remove technical noise from the client-facing view.

Sharing Options The main sharing modalities described are:

• Live URL with password protection: Share a specific prototype or set of screens via a link. Clients can view and add comments directly.
• Notion embedding: Copy the full-screen link and embed it as an iframe inside a Notion document. The prototype becomes clickable within the proposal document itself, giving clients a seamless experience without leaving the proposal.
• HTML export: Single screens can be exported as HTML files.
• PDF and image export: For static presentations or print.
• React zip export: The full React project with all code, styling, and structure is packaged as a zip file for developer handoff.

The Comment and Review Layer Within UX Pilot, clients can leave comments on specific screens or prototypes. The canvas organises outputs into categories: single screens, prototypes, and flows. Agencies can delete, reorder, or curate what is visible before sharing to present a clean and intentional review experience.

Strategic Thinking: Pricing, Margins, and What AI Does to Agency Business Models

Do Not Reduce Your Rates Adam is direct on the pricing question. When AI compresses the time needed to produce design work, the temptation is to lower rates to reflect the reduced hours. He argues the opposite: use the time saved to produce better quality, more iterations, and more personalised outputs. Clients who hire consultants want to avoid the homogenisation that comes from everyone using the same AI tools. A consultant who leverages AI to produce faster and more tailored results can justify the same or higher rates.

The Homogenisation Risk As more designers use similar AI generation tools, more products and websites will start to look the same. They will share patterns, visual conventions, and layouts because they share the same generative models. This is an opportunity for consultants who invest in differentiation — through design system integration, unique prompting strategies, and professional curation of AI output — to stand out precisely because their work does not look like everyone else's.

The Role of Human Judgment The conversation returns repeatedly to the question of what remains irreplaceable. Adam's answer is consistent: judgment, taste, and decision-making. AI can generate 20 versions of a UI element in the time it previously took to sketch one. But deciding which version is right for this product, this user, and this business goal is still a human function. The tools surface options; the designer decides.

He uses an example from UX Pilot's own product development: they use the tool to generate concepts for their own interface changes. Generating 20 versions of the header is trivial.

Deciding which version is obvious and intuitive for users while also achieving a business goal is where real time is spent.

The PM and Designer Disagreement Problem

The interviewer raises the question of whether any AI tool can bridge the gap when a product manager and designer fundamentally disagree on direction. Adam's answer is nuanced. AI cannot resolve disagreement — and blindly following data without human judgment can cause paralysis rather than progress. What AI does change is the cost of contributing ideas visually. A PM no longer has to write a long document with screenshots and annotations to communicate a design idea. They can generate a rough visual of what they mean, which changes the nature of the conversation from interpretation to reaction.

Everyone can contribute to the design process now, which means more concepts, more involvement, and ultimately more material for the designer to curate toward the best solution.

Target Audience and Customer Insights

Who Is Using UX Pilot The primary user segments are:

• Designers at agencies
• Designers at product companies and SaaS teams
• Engineers who need to prototype quickly
• Solo founders and startup builders The Founder Churn Lesson Early in UX Pilot's life, founders and solo builders showed the highest conversion rates and the highest initial revenue. Adam assumed this was the primary market. The data told a different story: founders churn quickly because their design need is project-bound. Once they have a working prototype or MVP interface, they move to the next phase and no longer need a design tool. Professional designers at agencies and product teams, by contrast, use the tool daily or hourly. That sustained usage pattern drives retention and predictable revenue. The team shifted focus accordingly, though founders still represent a meaningful segment of the user base.

Onboarding and Practical Adoption for Agencies

Getting to First Use Adam does not give a specific time-to-value number but frames onboarding as relatively fast given the tool's chat and prompt-based interface. The first meaningful use case for most agencies is the discovery-to-prototype flow: take a call transcript, paste it in, generate initial screens, and share with the client or use in the proposal.

Visibility of the AI The interviewer asks whether clients see that AI was involved. Adam explains there are multiple modes: agencies can be fully transparent and share the canvas including prompts, or they can share only the outputs with prompts hidden. The agency controls the level of AI visibility in the client-facing experience.

Manual Editing Within the Tool UX Pilot also supports non-AI editing, similar to Webflow's visual editor. Users can move elements, adjust layers, switch colours, and save elements as reusable symbols. This means the tool does not require re-generating everything from scratch if a small manual adjustment is needed, and it supports a hybrid workflow where AI generates the foundation and designers refine it manually.

What Is Coming Next

Design System Integration as a Service Adam flags deeper design system integration as the current development priority. Many professional users — both on the agency side and the in-house product side — are working with products that have mature, existing design systems. These can be React component libraries hosted on GitHub or Figma design systems. UX Pilot is building the ability to ingest these systems, verify they work correctly, and customise the integration so that everything generated through the tool is on-brand from the first output, without manual colour or component substitution afterwards. For larger clients with ongoing design workflows, this means every AI-generated screen is production-ready in terms of brand compliance.

API Access and MCP UX Pilot already has API access. MCP (Model Context Protocol) support is described as coming soon. Some enterprise clients are already running fully agentic workflows: user insights arrive from one source, are piped through UX Pilot's API, generate designs programmatically, and push prototypes to clients or prospects without any manual intervention in the middle. The upcoming MCP integration will make these agentic pipelines easier to build and connect.

TV Advertising The interviewer mentions seeing UX Pilot commercials on television. Adam confirms this is real, calling it "maybe crazy" but not elaborating further on the strategy. This is an unusual channel for a B2B SaaS design tool and suggests significant marketing confidence and budget, though the session does not explore this in detail.

Where to Find Adam and UX Pilot

Product: UX Pilot (free trial available)

• Community: UX Pilot Slack channel for direct access to the team
• Sales inquiries: Available for agencies with specific use cases or proposals
• API access: Available now
• MCP: Coming soon

About the speaker Adam Fard Founder of UX Pilot

Adam Fard is founder of UX Pilot, an AI-assisted design and prototyping tool. He works with agencies on new workflows for selling and delivering design work faster.