Skip to content

Quick start#

Three things to set up, in Settings.

1. Tell Alfred about you#

Open Settings → You and either drop your CV as a PDF onto the import box, or paste your résumé into the field.

Importing reads the PDF's text layer, asks your configured provider to structure it, and shows you what it found before anything is filled in — name, headline, years, skills, target roles and the résumé itself. Nothing is saved until you press Save settings, so you can review and correct first.

Scans will not work

Only the text layer is read. A CV that is a photo or a scanned image has no text to extract, and Alfred says so rather than quietly producing an empty profile. Export a text PDF, or paste the text in.

This is the single biggest driver of output quality. Every judgement Alfred makes about a role is measured against this text, so the more concrete it is — real projects, real numbers, the actual stack — the more specific the analysis. A one-line summary produces one-line advice.

Fill in your target roles, locations and compensation target too; they feed the salary read and the positioning angle.

2. Connect a provider#

Open Settings → AI layer, pick a provider and press Test connection.

Paste an API key from the Anthropic Console, or set ANTHROPIC_API_KEY in the environment and leave the field blank.

If the claude CLI is installed and signed in, pick this and press Test connection — there is nothing to configure. Calls use your existing subscription rather than an API key.

Any OpenAI-compatible server. For Ollama:

Base URL   http://localhost:11434
Model      qwen2.5:14b

Everything stays on your machine.

Point Alfred at your own HTTP agent. See the agent API.

More on providers →

3. Add an application#

Press Add application, paste a job posting, and hit Extract fields. Alfred pulls out the company, title, location, seniority, salary and a cleaned-up description. Check them and save.

On the application page:

  1. Analyze this role — a fit score, skill coverage, your gaps, and how to position yourself.
  2. Build my prep plan — coding problems, concepts, system design prompts and behavioural stories, each with a rationale tying it to this posting.
  3. Draft the questions — the interview questions this loop will likely ask, with answers written in your voice for you to rewrite.

Then drag the card across the board as the process moves. Every move is logged on the timeline.

Optional: your mailbox#

Settings → Mailbox connects a read-only IMAP account so recruiter mail lands in Alfred's inbox, classified and matched to your pipeline. Mailbox →