Product Requirement Document

Overview

AI App Builder Evaluation — Mobile News App Project:

AI News Platform Team

Contributors: Charlyn Truong

Status: In Progress


Background


The rise of generative AI development tools has lowered the barrier to software creation, enabling faster prototyping and application development across technical skill levels. As the market grows more crowded, it becomes harder to evaluate which tools actually deliver on that promise — especially for designers and early-stage builders working without an engineering team. This project was created to answer that question directly: by building the same product across multiple AI app builders using the same prompt, the goal is to create a firsthand, structured comparison of how each tool performs in practice.

Problem


Designers, developers, and non-technical builders are increasingly turning to AI app builders to accelerate product development — but the landscape is fragmented and hard to navigate. Each platform promises fast, high-quality output with minimal effort, yet they differ slightly in how they interpret prompts, handle iteration, and perform within free tier constraints. Without a direct comparison, choosing the right tool becomes guesswork.


This project removes that guesswork by testing four leading AI app builders — Lovable, v0, Bolt, and Emergent — against the same brief, the same evaluation criteria, and the same iteration process. The goal is to give designers and early-stage builders a clear, experience-based reference for which tool performs best and under what conditions.

Objective / Goals


The goal of this project is not to determine which AI app builder is best, but to observe, document, and compare how each tool responds to the same conditions — giving designers and early-stage builders an honest, experience-based reference.

  • Evaluate how each AI app builder interprets and executes the same initial prompt without modification

  • Observe how each builder responds to iteration prompts — specifically how well it handles small details, corrections, and feature additions

  • Document the features each builder offers out of the box versus what requires prompting or is unavailable entirely

  • Compare each builder's output against the discovery build in Replit as a neutral reference point — not as a standard of better or worse, but as a consistent baseline

  • Identify what each builder does well and where it falls short so designers and builders can make an informed choice based on their specific needs

Success Metric


Core — evaluation is only valid if these are present

  • Interest/category selection screen is present and functional on first output

  • Feed landing page renders with article cards organized by category

  • Category tags are visible and associated with each article card

  • Article cards are tappable and attempt to open the article — whether the link is live or not

  • All iteration prompts are reflected in the final output in some form — tracked in the feature implementation table below

  • Builder generates a name and establishes a basic design system from the initial prompt (building a prototype ready to publish application)

  • Builder can publish or share a live preview of the project

  • Builder allows rollback or access to a previous version of the project

Bonus Points — noted where present, not penalized where absent

  • Article links connect to real, working sources

  • Builder suggests design alternatives or variations when prompted

  • Color system or visual identity is applied consistently across screens

Key Features and Scope

What are we building


Feature 1 — Interest Selection & Personalized Feed

Description: Users select preferred news categories during onboarding to personalize their feed

Goal: Establish user preference from the start so content feels relevant immediately

Use case:

First time user opens the app, selects Business and Tech, feed shows only those categories

Additional details: Categories should be visually distinct — icons, color coding, or both


Feature 2 — Article Cards with Summary

Description: Each article displays as a card with a headline and 2–3 sentence summary

Goal: Let users consume key information without committing to a full article

Use case: User scrolling feed reads the summary and decides whether to open the full article

Additional details: Card should be tappable — opens summary modal first, full article second


Feature 3 — Article Summary Modal

Description: Tapping an article opens an in-app overlay with key points before redirecting

Goal: Reduce friction between scanning and reading — give users a middle step

Use case: User taps article, modal appears with 3 bullet points, user decides to read more or dismiss

Additional details: Modal should scroll independently, exit button clearly separated from read more button


Feature 4 — Engagement Actions

Description: Upvote, save, share, and comment options on each article

Goal: Give users ways to interact beyond passive reading

Use case: User finds an article worth sharing, taps share without opening the full article

Additional details: Saved articles should be accessible from the profile page


Feature 5 — Navigation & Trending

Description: Bottom navigation with Home, Trending, and Profile tabs

Goal: Give users clear wayfinding and a reason to return to the app

Use case: User taps Trending to see what articles are gaining traction from other readers

Additional details: Trending ranked by read count, votes, and active discussion


What we are not building

  • Real news API integration — all content is placeholder for evaluation purposes

  • Authentication or user accounts — profile is lightweight, no login required

  • Backend infrastructure — no database, no server, no data persistence across sessions

  • Cross-platform compatibility testing — mobile first, web view secondary

  • Production deployment — this is an evaluation prototype, not a shippable product

App Requirements / Build Targets

These are the features each builder will be prompted to produce. All builders receive the same starting prompt. The core list reflects the evolved target after iteration — the starting prompt, was MVP only.


Core features that are required and iterated across all builders

Necessary Features (prompted to add)


  • Interest Selection - Onboarding screen to pick topics and personalize feed

  • Scrollable Feed - Headline + couple of sentence snippet; tapping opens short description summarizing article - no redirection; comment section

  • Article Summary Modal - Key points shown before committing to full article

  • Read Indicator - Visual marker on articles already viewed

  • Bottom Navigation - Home / Trending / Profile tabs

  • Upvote / Save / Share - Pre-article actions; saved articles accessible in profile

  • Commenting - View and post discussion within modal - no full article scroll required

  • Trending Page - Surfaces articles gaining traction by read count, votes, and active discussion

Nice to Have

(what builder tools have to offer additionally)


  • Expo Go prototype preview (live app preview on phone)

  • Publish with live link through site

  • Real article links

  • Interchangeable between web view and mobile view

  • Organized time machine to return back to previous versions

Evaluation Criteria

Dimension
Rating
Assessment
01
Output Quality
Named the app, strong design language from prompt one, set up own design system
02
Ease of Use
Browser-based, no setup, no coding necessary, outputs working prototype
03
Iteration Handling
Good on additive changes, how well tool can understand an amateur prompt, can take described idea and implement, simple prompt vs complex prompt
04
Free Tier Limits
Limited tokens usage, publishing project, live previews / sharing project

🧠 Out of Scope

Real news API · Authentication · Backend logic · Production Deployment · Cross-builder feature parity