AI coding assistants are becoming part of everyday development. But there's still a common problem: different AI models, separate subscriptions, API keys, and tools that don't always work with your actual project.
That's why I built Watta — an AI Coding Agent for VS Code that brings 300+ AI models into one coding workflow.
Instead of switching between multiple AI tools, you can work directly with your codebase, choose a model, and let the agent help you build, debug, and refactor code.
What can Watta actually do?
Watta is more than a chat window inside your editor. It's a project-aware AI agent that can work with your existing files.
Here's what you can do:
- AI Coding Agent — describe a task and let Watta inspect your project, search files, and make code changes.
- Ask, Plan & Agent modes — understand code, plan improvements, or execute multi-step development tasks.
- Multi-file editing — make coordinated changes across multiple files.
- Code autocomplete — get inline code suggestions while typing, with supported access.
- Review and undo changes — inspect code changes and accept or revert them.
- Terminal integration — let the agent request commands, with approval controls.
- 300+ AI models — access a growing catalog including models from OpenAI, Anthropic, Google, DeepSeek, and others.
- One account — no need to configure separate API keys for every AI provider.
One extension, multiple AI models
One of Watta's biggest advantages is flexibility.
Different coding tasks may benefit from different AI models. Sometimes you need fast code generation. Other times, you want stronger reasoning for complex debugging or architecture decisions.
With Watta, you can explore supported models from one interface rather than installing a new extension for every provider.
You can also use Watta's own model options, including Watta Free, Lite, Pro, and Max, depending on your account access.
Try paid AI models for free
You can create a free Watta account and get started without connecting your own provider API keys.
We're also offering temporary promotional access to 3 paid AI models.
The selection rotates over time, giving developers a chance to experiment with different models.
Look for the Temp access label in the model picker. Available models and usage limits may change, so check the current offer before starting.
Let's try a real coding task
Imagine you have an existing project and want to improve a feature.
Open your project in VS Code and ask Watta:
Find the reason this form is not submitting correctly. Check the related files, explain the problem, and implement a fix. Let me review the changes.
The agent can inspect the relevant workspace files, investigate the issue, and propose or make changes directly in the editor.
You can review the result, accept the changes, or revert them.
You can also start in Plan mode if you prefer to review an implementation approach before editing.
Getting started takes about a minute
- Install Watta: AI Coding Agent from the VS Code Marketplace.
- Open the Watta panel in VS Code.
- Sign in with your free Watta account.
- Open your coding project.
- Choose an available AI model and describe your task.
That's it. No separate provider API keys required.
Why I built Watta
The idea behind Watta is simple: developers should be able to work with AI without managing a different tool, account, and integration for every model.
Watta is designed as a unified AI workspace, connecting coding workflows with AI chat and a single model catalog.
The VS Code extension is an important part of that experience.
We're actively improving it, adding features, refining agent workflows, and listening to developer feedback.
Try Watta and share your feedback
Watta is still growing, and early feedback from developers is particularly valuable.
If you're using VS Code and want to experiment with an AI coding agent that supports multiple models, give Watta a try.
Install Watta AI Coding Agent — VS Code Marketplace
What matters most to you in an AI coding assistant: model choice, code quality, autocomplete, speed, or agent capabilities?
I'd be interested to hear your experience in the comments.
Top comments (0)