Frequently asked questions
Everything you might want to know before hiring your first AI junior developer.
What exactly is an "AI junior developer"?
It's a Claude-powered agent that works in your codebase like a contractor — building features, fixing bugs, writing tests, and handling maintenance. Unlike an autocomplete tool, it takes a whole task, works in your repo, and opens a pull request with its changes. Unlike a fully autonomous agent, every change waits for your approval.
Why "junior" — is it not very good?
It's genuinely capable — it handles a large share of everyday development work quickly. We call it a junior because, like any junior, its work should be reviewed before it ships. That review step is a feature: it keeps you in control and means a confident-but-wrong change never lands silently in production.
Do I own the code it writes?
Yes — completely. Every line your AI junior produces is yours, with no licensing strings attached.
How is it priced?
Hourly pay-as-you-go, or monthly plans starting at $599/mo (part-time) and $1,999/mo (full-time). Commit to a 3-, 6-, or 12-month term and save 20–40%. Compare that to a human contractor at $80–150/hr — see the pricing page and savings calculator.
Which tools does it integrate with?
GitHub (core), plus Slack, Linear, Jira, Notion, and email. Assign work where your team already works, and get pull requests and standup-style summaries back.
Is my code and data secure?
Access is scoped to what you grant, secrets are kept out of the agent's workspace, and higher tiers add audit logs, a security review, and SSO. You decide what the junior can see and touch.
What can it actually do?
New features, bug fixes, refactors, test writing, dependency upgrades, integrations, internal tools, and ongoing maintenance across web, backend, mobile, and DevOps stacks. The free repo analyzer recommends the right specialty for your project.
What can't it do (yet)?
It's not a replacement for a senior architect on novel, high-risk system design, and it shouldn't merge to production unreviewed — that's the human-in-the-loop step. For most SMB workloads, that's exactly the right division of labor.
What happens if it gets something wrong?
You'll see it in the pull request before anything ships. Request changes and it iterates. Because nothing merges without your approval, mistakes get caught at review — not in production.
How do I start?
Run a free repo analysis to see your recommended junior, then pick a plan. No credit card to try.