Match the work, not just the title.
I refined the brief around website production, e-commerce, practical business tools, and hospitality. Required expertise, preferred qualifications, and unknowns need different treatment.
Job Desk / AI-assisted workflow automation
The job search
became the project.
I built a private workspace with AI assistance to find relevant opportunities, keep the evidence together, and turn the next step into something manageable.
01 / Research
Find the evidence.
Official sources
Saved search criteria
Open questions
02 / Review
Make a decision.
Relevant experience
Specific skill gaps
Application history
03 / Prepare
Take the next step.
Tailored drafts
Personal review
Manual submission
Workflow overview. The live workspace and its application records are private.
Less repetition. Better context.
A useful job search involves much more than collecting links. I was repeatedly comparing requirements with my experience, checking work arrangements, revisiting employers, and trying to remember what I had already reviewed.
I wanted a workspace that held those decisions together. The goal was to reduce repeated work while keeping the judgment about fit, accuracy, and applying with me.
From scattered listings to a clear next step.
- 01
Define a useful match.
I translated my experience, location, compensation needs, and preferred work into saved criteria. The brief distinguishes demonstrated skills from transferable experience and gaps, so a similar job title is not enough.
- 02
Research with evidence.
Scheduled AI-assisted research checks employer and applicant-tracking pages. The workspace retains source links, dates checked, fit explanations, and unresolved questions. A search date is kept separate from a posting date.
- 03
Keep a durable record.
Opportunities are deduplicated by source URL. The private app stores jobs, personal notes, application status, and research coverage, so a new search can build on earlier decisions.
- 04
Prepare, then review.
A saved profile supplies the facts for tailored résumé and cover-letter drafts. I review the format and wording, download the files, and decide what to submit. A download never counts as an application.
The decisions that shaped it.
Make uncertainty visible.
A listing can be promising without confirming Oklahoma eligibility, benefits, or travel. Those questions stay attached to the opportunity. The app also records which sources were actually checked.
Keep automation accountable.
Research can add context and document drafting can reduce blank-page work. Neither should invent experience, mark a job applied, or contact an employer without my decision.
A tool that found a relevant opportunity.
Job Desk surfaced an automation role asking for the same kind of work I was doing: noticing manual friction, defining a better process, and building a working tool with AI assistance. The search itself became a concrete example I could discuss.
Using the app also exposes the less visible work of ownership. A document-preparation failure became a debugging task involving output validation, useful error messages, and checking the complete path from a saved profile to a downloadable PDF.
Where it stands
A working private tool for my own search, with continuing refinement. I have not measured an aggregate time saving or claimed a hiring outcome. The system does not submit applications or contact employers.
How I used AI.
I used ChatGPT and Codex to help build and refine the application, research workflow, and document generation. My contribution is defining the problem, supplying accurate experience, specifying rules and exceptions, evaluating results, and directing changes.
Scheduled AI-assisted research and the app work together: the research checks opportunities, while the app holds the profile, evidence, decisions, and files. This is an independent AI-assisted project, not a claim of professional software-engineering employment.
What I bring to the next project.
I start with a recurring problem I understand, turn it into explicit requirements, and keep working through the exceptions that appear in real use. A useful tool needs both a clear workflow and someone willing to maintain it.
- Workflow design
- Requirements
- AI-assisted building
- Evidence-based review
- Iteration & troubleshooting