AI Assistant & AI Office
Two modules in one: an AI assistant integrated in the editor (copilot) and AI Office, a team of autonomous agents that plans and executes complex tasks autonomously.
AI Assistant (copilot)
Quick actions
Select any code snippet and access quick actions by tapping the AI button:
In-editor chat
Open the side chat panel to ask questions about the current file's code. The file context is automatically included in each message, so the model understands exactly what you're talking about without you needing to copy and paste code.
Available providers (4)
Google Gemini — requires your own Gemini API key or OAuth login via the Gemini CLI (Google AI Studio's free tier is generous: 15 req/min, 1,500 req/day). Pocket Code's Free plan does not add limits beyond what Google allows on your account.
Pocket Code does not manage or proxy any AI service. The app connects your device directly to the provider using your own key.
With the Premium plan the full provider list is unlocked. In every case the model is BYOK (you add your own API key) or OAuth (Claude Code, Codex and Gemini CLI). Pocket Code does not add credits: usage is billed by the provider directly to you.
Google Gemini
Free · BYOK / OAuth
Gemini 2.5 Pro · 2.5 Flash · 2.0 Flash · Flash Lite
OpenAI (Codex)
Premium · BYOK / OAuth
GPT-4.1 · GPT-4.1 Mini · GPT-4o · GPT-4o Mini · o3-mini
Anthropic Claude
Premium · BYOK / OAuth
Sonnet 4 · Opus 4 · 3.5 Sonnet · 3.5 Haiku
GitHub Copilot
Premium · OAuth
13 multi-vendor models (OpenAI, Anthropic, Meta Llama 4, Mistral, DeepSeek)
Four providers total. Copilot acts as a bridge to 13 additional multi-vendor models (Meta Llama, Mistral, DeepSeek…) without needing extra tokens.
AI OfficePremium
AI Office is a team of AI agents that work in coordination to execute complex development tasks: from designing the architecture to writing the code, tests and deploying.
How it works
- 1
Describe the task
Write what you want to build. Example: "Create a REST API with Node.js and Express for a JWT authentication system".
- 2
AI generates the plan
The Planner agent divides the task into phases, sprints and concrete tasks. Each task is assigned to the specialized agent.
- 3
Review and approve
Review the plan before executing. You can modify, delete or add tasks.
- 4
Monitor execution
Follow the progress in real time in the task board. Agents write code, create files and run commands.
- 5
Review the result
When finished, review all created files and executed tests.
Agent roles
🧠 Planner
Analyzes the task, designs the architecture and divides work into sprints
⚙️ Backend
Writes APIs, business logic, models and database migrations
🎨 Frontend
Creates UI components, styles and client-side logic
🧪 QA
Writes unit and integration tests and verifies coverage
Next
Workflows