NewsBite
made by Bolt
The initial prompt resulted in error
Bolt was the most technically unstable builder in this evaluation.
The initial prompt was submitted twice across two separate projects — both times the builder reported successful completion in the chat while the preview returned uncaught errors. To isolate whether the issue was the builder or the prompt, a simpler unrelated prototype was tested next and built without any issues. The original prompt was then resubmitted in a simplified form to reduce complexity. The result was the same error.
Simplifying the prompt for easier implementation
The final attempt prompt:
"Build me a mobile application for on the go users to stay up to date with current news. Have the first page be categories that the user can pick from for preference on what they want to read about. The main page should be article cards with a summarize couple of sentences of what the article is about. Make the UI clean and minimalistic, easy for the user to read. There should be a button for users to select to be redirected to the article."
Bolt seemed to need more than the simple prompt. Adding the first iteration from Lovable, Bolt was able to produce a working prototype.
Bolt seems to need the right amount of information and context to get an idea of what to build. This may require more than one prompt or less detailed prompt from an AI agent to produce something on here that works.
Next prompts building towards final
The second iteration focused on expanding engagement and navigation — adding a share option, a read tag indicator, and a commenting system per article. The third iteration went deeper into refinement: full articles opening within the app with a back arrow to return to the feed, navigation restructured to move saved and upvoted articles into the profile page, topics condensed into a filter button on the home screen, article cards resized, spacing adjusted between action buttons, and a personalized greeting added above the feed. Bolt handled the structural changes reasonably well — the navigation restructure and in-app article view came through cleanly — but several smaller details required correction and not all of them resolved fully.
A persistent issue
The most persistent issue across iterations was the comment feature within the article modal. When the comment button was selected inside the modal, the system pulled the user out of the modal entirely — reverting the background to the home feed while the comment section attempted to load. A correction prompt brought the user back to the article after closing comments, but the underlying problem remained: the background continued showing the feed while the comment modal was open, breaking the visual continuity of the experience. This is the kind of small but noticeable UX detail that Bolt struggled to fully resolve through prompting alone — it understood the ask but couldn't execute the layering correctly. It reflects a broader pattern observed throughout the Bolt evaluation: structural changes landed, fine-grained interaction details did not.
Bolt pricing
Key takeaways
Bolt had the rockiest start of all four builders — multiple failed attempts, error codes, and missing project files, all while the chat reported successful completion. That disconnect between what the builder communicated and what it actually delivered was the most disorienting part of the experience. What eventually unlocked Bolt was more context upfront — once given enough direction it produced a working prototype and handled structural changes reasonably well across iterations. Where it consistently fell short was in fine-grained interaction details, the comment modal layering being the clearest example — understood in concept but never fully executed through prompting alone. Bolt is a builder that rewards having done the design process beforehand and suits a user who prefers building through conversation rather than a single structured prompt. The free tier complexity ceiling is a real limitation, but with more context, more patience, and clearer direction going in, Bolt has potential — it just needs more from the user than the other builders to get there.