Frequently asked questions

Does Sofik replace my IDE?

No. Sofik lives alongside it. You stay in your editor; when you need an agent, you trigger the HUD, it works in parallel. Your PRs stay in your normal flow.

Can I use my own key (Claude/OpenAI)?

Yes. Each agent in your org has its own config — model, key, prompt. No proxy in the middle skimming margin.

Where does my data live?

Sessions and memory live in a Postgres per org. On the managed edge, that’s our Postgres with tenant isolation. On self-host, it’s yours.

Does it work offline?

The desktop app does — the window opens, history is visible. Agent invocations need internet to talk to the LLM. It syncs back when connectivity returns.

How does multi-agent work?

Each agent is a persona with role, skills, model, and base prompt. They can delegate to each other (CEO → CTO → coder) with cancellation and telemetry propagated. You see the flow in the panel.

Support?

Email with 24h SLA (hi@sofik.ai), shared Slack for paid accounts.

Privacy?

We don’t train models on your conversations. We don’t sell data. LLM keys never pass through our servers — every call leaves your machine directly.