The Best Vibe Coding Tools in 2025
By Greg Isenberg
Key Concepts
- Vibe Coding Apps: Tools that allow users to generate code or applications through natural language prompts, often with AI assistance.
- Tier List: A ranking system used to categorize and compare different vibe coding apps based on their features, usability, and perceived quality.
- Technical vs. Non-Technical Users: The distinction between users with coding knowledge and those without, influencing the suitability and ease of use of different tools.
- AI Models: The underlying large language models (LLMs) that power vibe coding apps, such as Claude and GPT.
- Agent/Tooling: The additional logic and capabilities built around an LLM that enable it to interact with files, perform actions, and generate more complex outputs.
- Community and Ecosystem: The importance of a strong community, documentation, and available integrations for the success and adoption of a development tool.
- Backend: The server-side infrastructure and logic that supports an application, crucial for many vibe coding platforms.
- Mobile App Development: The trend and opportunity in building mobile applications using AI-powered tools.
- Mindset Shift: The necessary change in perspective for non-technical users regarding the time, effort, and iterative process involved in software development.
Tier List and App Analysis
This episode presents a "definitive guide" and "ultimate tier list" of vibe coding apps, offering a non-professional, yet informed, opinion on their strengths and weaknesses. The discussion categorizes apps into tiers, with a focus on technical requirements, user experience, and underlying technology.
D Tier
- Windsurf:
- Reasoning: Placed in D tier primarily due to trust issues stemming from the founder's departure and the subsequent acquisition by Devon. While the technology itself might be good, the team's actions have eroded confidence.
- Technicality: Described as "very technical," requiring a deep understanding to maximize its potential.
- Developer Focus: The team explicitly states they are building for developers, not non-technical users.
- Personal Stance: Despite acknowledging its potential, the speaker would "never use Windsurf ever again" due to the business situation. Another speaker would have rated it A or B tier based on past experience but acknowledges the business concerns.
A Tier / S+ Tier
- Cursor:
- Reasoning: Highly regarded, with one speaker placing it in S+ tier and the other in A tier.
- Technicality: Requires technical proficiency to use, involving actions like forking repos and running npm commands. Non-technical users will likely experience "pain."
- Community and Tutorials: Benefits from a large community and abundant tutorials, similar to the advantages of frameworks like Next.js or libraries like React. This strong ecosystem is a significant pro.
- Developer Focus: Explicitly built for developers, though non-developers can use it with effort.
- Key Feature: Mention of "plan mode" as a useful feature for planning.
- AI Model: Utilizes Claude Sonnet 4.5 as its preferred coding model.
B Tier / A-B Tier
-
Lovable:
- Reasoning: Placed in B tier. The primary pain point addressed by Lovable is the integration complexity of previous versions (Lovable and Bolt), which required backends and API key management. Lovable abstracts these away.
- Ease of Use: Aims to simplify the process by automatically handling API calls upon project boot-up.
- Comparison: Despite improvements, some users still gravitate towards tools like Cloud Code, Codeex, and Cursor.
- Backend Limitation: Lovable dictates the backend choice, which has pros and cons depending on the app's needs.
-
Bolt:
- Reasoning: Placed on par with Lovable, in the B tier.
- Investment Note: One speaker is an investor in Bolt.
-
V0 (Vercel):
- Reasoning: Placed in an "in-between" tier, close to A tier, and considered highly suitable for "vibe coders" (non-technical users).
- Technicality: Described as "non-technical friendly."
- Vercel Integration: Benefits from Vercel's marketplace and integrations, offering a "plethora of options" for backends, unlike Lovable's more curated approach.
- Future Outlook: Seen as having a strong future due to its tight integration with Vercel.
- Real-world Application: Successfully used for an RSVP website when other AI tools failed.
- Features: Offers reusable templates and components.
- Bias Disclosure: One speaker works for Convex, which will be on the Vercel marketplace, acknowledging a potential bias but stating they liked the product before joining.
-
ROR (Mobile App Vibe Coding):
- Reasoning: Placed in B tier, alongside Lovable and Bolt.
- Focus: Specializes in building mobile apps.
- Technology: Believed to be built on Expo for cross-platform compatibility (iOS and Android).
- Newness: Relatively new, launched a few months prior to the recording.
-
Vibe Code App (Mobile App Vibe Coding):
- Reasoning: Placed in B tier.
- Analogy: Described as the "Lovable for mobile apps."
- Focus: Mobile app development.
- Technology: Believed to be built on Expo.
- Newness: Relatively new.
-
Anything (Mobile App Vibe Coding):
- Reasoning: Briefly mentioned, with limited personal experience.
- Focus: Mobile app development.
- Newness: Part of the emerging trend of mobile app vibe coding platforms.
-
Codeex:
- Reasoning: Placed "a little higher" than Lovable and Bolt, but below Cursor.
- Hype: Gaining significant hype, particularly with its GPT-5 model.
- Performance: Described as "okay" and "pretty good" when used within Cursor, but Claude Sonnet 4.5 is considered superior.
- Non-Technical Use: Has a web version and is accessible to non-technical users, with many tutorials available.
- Improvement: Recognized for significant improvement ("most improved award") from its initial state.
- Developer: Sam Altman is credited with its development ("Sam's cooking").
-
Replet:
- Reasoning: Placed in the same tier as Lovable and Bolt (B tier).
- Usage Mystery: Acknowledged that many people don't know anyone who uses Replet, yet it's "winning" and has raised significant funding.
- Modes: Offers both a "vibe code mode" and an "agent mode" where it can build an app in 15-20 minutes.
- Output: Described as "pretty impressive."
- Backend: It's unclear if they are developing their own models or using existing ones.
-
Chef:
- Reasoning: Placed in the same tier as Lovable, Bolt, and Replet (B tier).
- Purpose: Built by Convex to showcase the benefits of using Convex as a backend for vibe coding platforms, not to compete directly.
- Features: Offers immediate output and is open-source.
- Limitations: Lacks features like domain assignment, as it was not designed to impress end-users but rather developers building tools.
Cloud Code
- Reasoning: Previously considered S tier, but has been "nerfed" and is now ranked below Cursor.
- AI Model: Uses the same Claude Sonnet 4.5 model as Cursor.
- Agent/Tooling Difference: The key differentiator is the "agent" built on top of the LLM. Cursor's agent is perceived to be better at reading, writing, and editing code, giving it an edge over Cloud Code despite using the same underlying model.
- Future Outlook: Expected to be a "tug-of-war" between Cloud Code and Cursor, with their rankings potentially fluctuating.
- Non-Technical User Recommendation: If the speaker were non-technical, Cursor would be preferred over Cloud Code.
Mobile App Development Trend
The discussion highlights a significant trend in building mobile apps using vibe coding platforms.
- Opportunity: There's a perceived opportunity to make money by building apps, similar to the success of apps like Calis.
- Consumer Behavior: TikTok is used as a search engine, leading to impulse downloads of apps that address immediate needs or interests.
- Commitment Perception: While downloading an app feels like a commitment, targeted advertising and immediate value propositions can drive downloads.
- Technology: Many of these mobile app platforms are built on Expo to ensure cross-platform compatibility.
Advice for Users
For Technical People:
- Recommendation: Cursor or Cloud Code are the primary recommendations.
- Reasoning: Technical users are likely to be "steering the wheel" and not purely relying on "vibe coding." They will be more involved in the development process.
- Other Tools: Open Code by Dax is mentioned as a tool highly regarded by developers, though not personally tested by the speaker.
For Non-Technical People:
- Mindset Shift Required: A crucial change in perspective is needed. Building functional software that people will use and pay for takes time, planning, and iteration, not just a few prompts.
- Audacity and Care: Treat the process with respect. Expect initial versions to fail and be prepared to try different tools and approaches.
- Software is Difficult: Understand that while websites and landing pages might be easy, building robust software is challenging.
- Iterative Process: Embrace the learning process, where mistakes lead to experience and better building skills.
- Tools and Resources: Utilize tutorials, planning tools (like Cursor's plan mode), and brainstorming with AI (like ChatGPT 5).
- Patience and Enjoyment: The process of building something awesome takes time and should be enjoyed.
Betting on Founders
A final piece of advice is to follow the founders of these vibe coding companies.
- Reasoning: Users are essentially "betting on someone" and a company. Connecting with a founder's vision and belief can be a strong indicator of a platform's future success.
- Example: The negative experience with Windsurf is cited as an example of what happens when you don't bet on the right team.
Conclusion
The episode provides a comprehensive overview and ranking of various vibe coding applications, emphasizing the technical requirements, user experience, and underlying technologies. While tools like Cursor are highly praised for their power and ecosystem, the discussion also highlights the emerging trend of mobile app development and the critical mindset shift required for non-technical users to succeed in the realm of software creation. The importance of community, founder vision, and the iterative nature of development are recurring themes.
Chat with this Video
AI-PoweredLoad the transcript when you're ready to chat so the initial page stays lighter.
Related Videos

India Can Create The Largest AI Companies
Y Combinator

AI Boom in London | Bloomberg Tech: Europe 6/12/2026
Bloomberg Technology

Lachlan Harris warns Labor's budget isolates voters amid One Nation rise | Insiders on Background
ABC News In-depth

Tech sector warns tax could "double" under CGT changes | Insiders on Background
ABC News In-depth

YC Paper Club
Unknown Author

Making $$ with AI Agents
Greg Isenberg

NYC Innovates is BACK! #tech #innovation
Cheddar