Sadly through my very mixed experience in the past week, it is time to start thinking about life outside Replit. From the sentiment on this forum, I assume many are thinking the same.
Let’s compile alternative ideas, to support each other if we decide to pull the plug. Please add your thoughts below.
Personally, I will give Replit a little more time to fix things, mainly because my AI Dev Kit is designed around Replit. I believe they need a chance to right the ship. And perhaps, once we have heard others’ experiences elsewhere, we may realise Replit is no worse than others.
So what other platforms, tools and ideas do people have? What have you used or played with - good and bad?
I think it is fair to say AI is moving so fast, there is no perfect answer and all tools will go through growing pains. But hearing everyones experiences will be massively helpful to the group.
“We’re hearing that Agent 3 is too aggressive about fixing technical debt, which can overwhelm large, older projects. We’re addressing this. For new projects, Agent 3 lays a strong foundation and refactors continuously, enabling apps to go much further than current LLMs—essential to move beyond prototypes and throwaway code. For legacy projects, rapid debt reduction can be slow and costly, so we’ll dial it back. We’ll also add more controls over time. Expect an update this week.”
“Expect an update this week” is key. While they work on fixes, I’m testing Firebase Studio as an alternative in case Replit doesn’t adjust Agent 3’s pricing and bugs. Even if Agent 3 improves, my main concern is pricing. I’ve spent a lot over the past two weeks, and it doesn’t seem sustainable.
I tried Lovable, Bold, and Buble, but I’m looking for a server-side solution like Replit. After three weeks of using replit (14 hours a day), I’m still impressed but also disappointed. I joined Replit at a bad time—if I had started today or even next week after they fix the initial Agent 3 issues, I’d be far more impressed.
Briefly: Firebase Studio is a cloud-based, agentic development environment powered by Gemini, offering everything needed to quickly build and ship production-grade AI apps in one place.
Coding workspaces also enable you to:
Simplify coding workflows: Write code and test features, all with assistance from Gemini in Firebase at every step of the way. Complete a variety of tasks like debugging, testing, refactoring, explaining, and documenting code with ease.
Enhance existing apps: Import existing codebases from your local machine or git-based repositories, including GitHub, GitLab, and Bitbucket. Create custom templates for your preferred tech stacks to share across your team.
Create full-stack experiences: Customize and evolve all aspects of your apps, from AI model inference, agents, and retrieval augmented generation (RAG), to the user experience, business logic, database, and more. Easily expose and integrate tools such as APIs and microservices to your AI apps.
Work with familiar tools: Bring along your specific configurations, such as system tools, extensions, and environment variables to tailor your workspaces. Access thousands of extensions from the Open VSX Registry.
Flexible deployment options: Set up your app to run on the cloud with built-in integrations to Firebase backend services and Google Cloud Run. You can also deploy on your own custom infrastructure.
What do you think about fibase studio, anyone tested successfully?
@danielgadus This is very encouraging. He seems responsive to his twitter feed. I think I might pay more attention to his posts. Thank you for sharing it.
I really hope they address this, it put a wrench in the gears of development. I’m averaging about $1000-$1200 a month in usage and I actually went on vacation for a week this month, I assume I would have racked up who knows how much, I returned with agent 3 launching. Overall messy launch, they should credit everyone anything charged after the launch
I don’t fully understand, there is an alternative available already.
Replit+Assistant ONLY. 5 cents a call and have it follow tasks.
A bit more micro-management but at least i am in control and is working fine so far for my tasks.
I just tried to have Agent fix something, 3 times - nothing. I asked assistant to find the problem first (important step) and then after it fixed it. Simple.
Assistant is still very good. It can mess things up sometimes, but easy to rollback. This is a great workaround for now if you’re struggling with Agent 3 usage on Agent 2 started projects.
I was really happy with Replit, despite spending half my time fixing errors, but now with version 3 I’m a bit panicked. Some things that were working suddenly stopped working (user login), then a few days ago I tried unsuccessfully to activate an accurate PDF printout, one failed attempt after another (after half an hour of trying, the architect realised that the layout required a different approach), so I had to stop because suddenly there was a cost I had never seen before.
So I’m sorry, I’ll wait because I don’t have unlimited funds, but I’m looking for alternatives and the one I tried recently looks very promising. It’s called Solid - trysolid.com - It doesn’t have all the features of Replit, but I think it’s very interesting.
No, there is an actual “assistant” that you can open in a new tab - it’s a very basic code assistant and inexpensive to run. But, it also can muck things up. Beware
