- Published on
Getting a free AI key for JobHakken (BYOK, step by step)
- Authors
- Name
- The JobHakken Team
- @jobhakken
JobHakken's AI features — résumé tailoring, cover letters, ATS keyword analysis, the Copilot — need a language model to run. You have two ways to power them:
- Bring your own key (BYOK) — free to set up, you pay the model provider directly for what you use (often pennies per action, or nothing at all with a local model).
- Managed — JobHakken runs the calls for you through a metered, zero-retention proxy, included on paid plans.
This guide covers BYOK, since it's the free path and the one most people start with.
Where the key goes
Open the desktop app → Settings → Your AI key. There's a Your AI provider picker with nine options. Pick one, paste your key (unless it's a local runtime — those need no key at all), and every AI feature in the app switches to it immediately. Nothing else to configure.
The same key works in the browser extension — screening-question drafts there use whatever key you've set in the desktop app, synced over the local bridge. Nothing is ever sent to JobHakken's own servers on this path: the request goes straight from your machine to the provider you picked.
Picking a provider
If you're not sure where to start, OpenRouter is the easiest first key — one account, hundreds of models, and it's the app's default. If you already have an OpenAI, Anthropic, or Google account, using that provider directly means one fewer middleman.
| Provider | Get a key here | Notes |
|---|---|---|
| OpenRouter (default) | openrouter.ai/keys | One key, hundreds of models — the simplest starting point. |
| OpenAI | platform.openai.com/api-keys | Uses gpt-4o-mini by default (fast, cheap). |
| Claude (Anthropic) | console.anthropic.com/settings/keys | Uses claude-haiku-4-5 by default. |
| Gemini (Google) | aistudio.google.com/apikey | Google AI Studio issues Gemini keys for free — generous free tier. |
| GLM / Zhipu | open.bigmodel.cn | Often the cheapest per-token option. |
| Ollama (local, no key) | ollama.com | Runs a model entirely on your machine — nothing ever leaves it, not even to a provider. |
| LM Studio (local, no key) | lmstudio.ai | Same idea as Ollama, different local runtime. |
| Codex (local, no key) | — | If you already run a local OpenAI-compatible app-server. |
| Custom (OpenAI-compatible) | — | Point it at any other OpenAI-compatible endpoint you run or have access to. |
Each provider's own page in the picker links straight to its key page, so you don't need to remember this table — it's the same list, in the app.
The truly free option: run it locally
If you'd rather not sign up for anything, Ollama or LM Studio run a model directly on your machine — select one of them in the provider picker, no key field to fill in, and no request ever leaves your computer. The trade-off is you need a reasonably capable machine and a model downloaded locally; cloud providers are faster to set up and usually produce stronger results from a smaller local footprint, but local means the complete privacy story — nothing crosses the network at all, even encrypted.
Cost
You're billed by the provider you chose, not by JobHakken — check their pricing page for exact rates. Most résumé/cover-letter/keyword-analysis actions use a small, bounded amount of text, so even a pay-as-you-go OpenRouter or OpenAI key typically costs a fraction of a cent per action. Set a spending cap in your provider's dashboard if you want a hard ceiling.
If you'd rather not set up a key at all
The Free plan works fully without one — job search, tracking, résumé building, and everything else that isn't AI-powered has no key requirement. AI features simply wait for you to add one, whenever you're ready.