A friendly AI assistant that helps you succeed with your projects.
The New User Assistant is designed to be helpful without being overwhelming. Unlike a raw AI tool that might race ahead with assumptions, this assistant:
- Asks before acting - Makes sure it understands what you want before starting work
- Shows its plan - Tells you what it's going to do so you can course-correct early
- Verifies its work - Actually tests things instead of just claiming they work
- Can be resumed - Keeps track of progress so you don't lose work if something interrupts
This bundle is ideal if you:
- Are new to working with AI assistants
- Want a more guided, careful experience
- Prefer the assistant to check in rather than assume
- Want confidence that the work is actually verified
If you're an experienced user who knows exactly what you want and prefers speed over guidance, you might find this bundle too cautious. That's okay - it's not for everyone.
Just talk to the assistant like you normally would. If your request is clear, it will proceed. If something is unclear, it will ask:
You: I need a script to process CSV files
Assistant: I'd like to make sure I build the right thing. A few questions:
- What do you want to do with the CSV files? (filter, transform, merge?)
- Where are the files coming from?
- What should the output look like?
If you want the assistant to just figure it out, you can say so:
You: Just do your best, I trust you
Assistant: I'll proceed with these assumptions: [lists assumptions]. Say "proceed anyway" if this works for you.
You: Proceed anyway
The assistant will then start work with its best interpretation.
For bigger projects, you can use the guided-start recipe for formal checkpoints:
amplifier recipes execute new-user-assistant:recipes/guided-start.yaml \
--context '{"initial_request": "Build me a task management app"}'This will walk you through:
- Clarify - Gather requirements (approval required)
- Plan - Design the approach (approval required)
- Build - Implement and verify (approval required)
This assistant is not a replacement for your judgment. It will:
- Ask questions, but you decide the answers
- Show plans, but you approve them
- Verify its work, but you do the final check
- Make recommendations, but you make the decisions
It's a careful helper, not an autonomous agent.
- Be specific when you can - "Add login with Google OAuth" is better than "add login"
- It's okay to not know - If you're not sure what you want, say so. The assistant will help you figure it out.
- Check the verification - When the assistant says "verified," look at the evidence it provides
- Use the override sparingly - The questions are there to help you get better results
If the assistant seems stuck or confused:
- Try rephrasing your request
- Answer its questions even if they seem obvious
- Say "let's start over" to reset the conversation
This bundle runs locally through Amplifier. Your conversations and code stay on your machine.