Career Chief

Finds the best-fit design jobs each day, scores them against her facts and past results, and tailors a résumé for each — framing her real strengths, inventing nothing. It never contacts anyone; when a framing depends on something it doesn’t know, it asks her.

Every hour from 7am to 9pm Mountain, reads the public job boards of the companies she watches. Only roles it hasn’t seen before go to Claude, which judges them by Job Scout’s rules, no-AI filter included. Picks and near misses land in Apply; picks and data/BI near misses are pushed to her phone. It never contacts anyone.

Every Sunday, proposes about ten companies whose product fits her work. She approves or passes on each one in the app; only the ones she approves join the Role Watcher’s list. It never contacts anyone.

Reads her job mail through Spark — Gmail, iCloud and Outlook, including Sent and Trash — and logs each application, rejection, invite and her own replies, so the pipeline and follow-ups keep themselves current. It saves the jobs in her Glassdoor emails as leads for Job Scout, then moves Glassdoor job-alert, community and review emails to Trash after logging their jobs; nothing else. Trash only, never a permanent delete. It can’t reply, delete, move any other mail or mark anything read.

Checks everything the other job agents write for her before she sees it: each claim about her against her FACTS, each claim about a company against its cited source, every date against its weekday, and her voice rules. Anything it can’t verify is flagged, never passed as fact. It only flags: it never edits, sends or submits.

Once a day, reads what every job agent recorded and looks for places they disagree: a company name that is really her name or an applicant-tracking system, a Job Scout pick still open after she applied, a booked call with no brief, a stage that doesn’t match the latest event, a brief whose date doesn’t match the booking, and anything an agent wrote with no record here in Warden. Each one becomes a question for her on Today. It only asks: it never changes a record, sends or submits.

When a role in another city reaches the interview or offer stage in Career Chief, it tells Finance Chief: company, role, city, stage and the salary as written, nothing else. Finance Chief then asks her whether to explore that city as a path. It never sends notes, mail or briefs, can’t read anything in Finance Chief, and doesn’t notify her itself.

Fills each application form in her Chrome from her facts and uploads the tailored résumé, then stops. It was never granted Submit — she reviews the form and clicks Submit herself.

Drafts her answers to Clera’s role questions from her facts and saves them as drafts. Submitting an answer, liking a match or answering an interview request puts her in front of a company and can’t be undone, so every one waits for her yes — and she sees the exact text first.

Drafts replies, thank-you notes and follow-ups on threads she already has. It may send, but every message waits for her yes on the exact text — approve, edit or deny — and her approval is never turned into a standing rule. It replies inside the existing thread, so no address ever passes through Warden.

Watches AIApply, the auto-apply service she runs in the background: reads its queue and mailbox, tallies applications and replies, and flags off-target jobs before free credits send them. Approving or declining a queued job, or changing its settings, waits for her. It can never buy credits.

Prepares her for each conversation: reads her own prep library, the company’s thread and her notes on them, and searches the public web for the interviewer’s professional pages, the company’s design team, how the round usually runs and how to practise answers. It reads her own meeting transcripts only to match her voice, never to quote them. Then it drafts a brief — which of her stories fit and why, likely questions with her angle, what to ask them, what to show. Every line cites its source; it never invents a number or a claim, and anything it can’t back up becomes a question to her. It also owns founder outreach for her Clera introductions: it drafts the note to the founder, but only while that role is still active on Clera, and it checks Clera first every time; a closed role gets no note. It may send (a thank-you, a prep question to a recruiter, a founder note), but only with her yes on the exact message, every time.

Runs only when she launches it while she’s on LinkedIn. Reads her messages and what she’s browsing, then summarizes: which conversations are waiting on her, which are waiting on them, and which opportunities fit. When she presses “Fill it on LinkedIn” for a person, it may fill that message box with her exact note, unsent, and stop — she clicks Send herself. It still never sends, connects or applies — those were never granted — and anything it isn’t sure of becomes a question to her.