He Built an AI Job Search Agent After Getting Laid Off — Now It Has 30K GitHub Stars
Mads walks through the open-source Claude Code job-search repo he built while unemployed after a December layoff: a profile builder, a multi-site job scraper, a scored fit evaluation, and a drafter/reviewer agent pair that writes grounded CVs and cover letters. It doubled as GitHub's number one repo of the week twice, and the interview doubles as an argument that packaging your workflow as shareable skills now beats writing a blog post about it.
The Next New Thing19 minTranscript found
Quick learning frame
Read this before watching.
AI strategy is choosing where agents create durable leverage, then managing scope, adoption, risk, and measurable outcomes.
New playlist item from The Next New Thing; queued for transcript-backed review, topic mapping, and a practical learning artifact.
Skill you build: The ability to turn a repetitive personal process into a packaged, shareable Claude skill set with a scoring gate and an adversarial reviewer agent, rather than a one-off prompt or a write-up of what you did.
Watch for the shift from claim to mechanism. The learning value is the point where the transcript reveals a repeatable action, tool boundary, context move, review habit, or artifact.
Concept diagram
Where this video fits.
01Use Case
02Workflow
03Agent Role
04Metric
05Risk
06Adoption
Deep lesson
Turn this video into working knowledge.
3,738 cleaned transcript words reviewed across 1,137 timed caption segments.
Thesis
He Built an AI Job Search Agent After Getting Laid Off — Now It Has 30K GitHub Stars teaches a practical ai strategy move: Mads walks through the open-source Claude Code job-search repo he built while unemployed after a December layoff: a profile builder, a multi-site job scraper, a scored fit evaluation, and a drafter/reviewer agent pair that writes grounded CVs and cover letters. It doubled as GitHub's number one repo of the week twice, and the interview doubles as an argument that packaging your workflow as shareable skills now beats writing a blog post about it.
The goal is not to remember the video. The goal is to extract the operating principle, tie it to timestamped evidence, test how far the claim transfers, and make something reusable.
0:30
Profile, scrape, score
“of the week twice. Now, we're going to do a screen share where he'll show you how this system automates a job hunt and how you can use it, too. But, the other takeaway that I want you...”
The workflow starts by feeding old CVs and cover letters in so Claude assembles a single ground-truth profile, then a job-scraper command sweeps whichever portals you configured (which sites matter varies by country) and returns titles, companies, locations, deadlines, and source URLs bucketed into high, medium, and low matches. A separate /apply evaluation scores a specific role across technical skills, experience, behavioral fit, location and commute constraints, and career alignment, producing a number like 73/100 plus explicit strengths and gaps. Assemble your own ground-truth profile document from every past CV and cover letter you have, then write the five evaluation dimensions you would actually want a role scored on before you look at his.
6:38
Reviewer with fresh context
“letter and your CV for for this specific role. And then it's sends that off to a reviewer agent which has a fresh context window, so no no information of who you are and it's it's not really...”
The drafter writes a role-specific CV and cover letter, then hands it to a reviewer agent running in a fresh context window with no knowledge of who you are, so it is unbiased when it scrutinizes every claim against the ground profile and asks whether Mads actually did this or whether it has been inflated. Mads got the idea from LLM-as-a-judge in RAG systems; his first draft-only version blew up his record, claiming publications he did not have, and results only became publishable once the critique loop existed. Take any document an agent wrote about you and rerun it through a second agent in a clean session whose only job is to flag unsupported claims against your source profile, then count how many it catches.
12:24
Ship the workflow
“I've been so happy about the open source community myself and all the things that that people put out there on on for example GitHub with the repositories and how they they share the knowledge. Um so, I...”
Publishing the repo (MIT, roughly 9,000 forks and approaching 30,000 stars, over 200 pull requests) turned out to be a better teaching artifact than a blog post because it hands people the tool that produced the result, not a description of it. Roughly half the repo is plain text instructions rather than code, and the contributed dashboard shows the community extending it; Mads deliberately left auto-apply out to keep a human in the loop, treating the fit score as a filter rather than playing a volume game (69 applications, about 21 interviews). Pick one workflow you have already refined for yourself, write it up as a skills directory with commands and instructions in plain text, and publish it under a permissive license instead of writing the blog post about it.
01
Use Case
Start with this video's job: Mads walks through the open-source Claude Code job-search repo he built while unemployed after a December layoff: a profile builder, a multi-site job scraper, a scored fit evaluation, and a drafter/reviewer agent pair that writes grounded CVs and cover letters. It doubled as GitHub's number one repo of the week twice, and the interview doubles as an argument that packaging your workflow as shareable skills now beats writing a blog post about it. Treat "Use Case" as the outcome you are trying to make visible, not a topic label. Anchor it to 0:30, where the video says: “of the week twice. Now, we're going to do a screen share where he'll show you how this system automates a job hunt and how you can use it, too. But, the other takeaway that I want you...”
02
Workflow
Use "Workflow" to locate the part of the ai strategy workflow the video is demonstrating. Ask what changes in your real setup if this claim is true. Anchor it to 6:38, where the video says: “letter and your CV for for this specific role. And then it's sends that off to a reviewer agent which has a fresh context window, so no no information of who you are and it's it's not really...”
03
Agent Role
Turn "Agent Role" into the reusable artifact for this lesson: A one-page business case for one agent workflow. This is where watching becomes something you can inspect and reuse.
04
Metric
Use "Metric" as the application surface. Decide whether the idea touches a browser flow, a local file, a model choice, a source document, a UI, or a review step.
05
Risk
Use "Risk" to prove the lesson. The evidence should connect back to the video title, transcript anchors, and a concrete output, not a generic best-practice claim.
06
Adoption
Use "Adoption" to carry the idea forward: save the prompt, checklist, diagram, or operating rule that would make the next agent run better.
Example
Source-backed work packet
Convert the video into a scoped task that includes the transcript claim, target workflow, acceptance criteria, and proof. The output should be a one-page business case for one agent workflow..
Example
Claim vs. demo brief
Separate what the speaker claims, what the demo actually proves, and what still needs outside verification before you adopt the workflow.
Example
Teach-back module
Transform the lesson into a definition, a mechanism diagram, one misconception, one practice exercise, and a check-for-understanding question.
Do not learn it wrong
Treating the title as the lesson without checking what the transcript actually says.
Letting the prompt drift into generic advice that could apply to any video in the playlist.
Copying the tool setup without identifying the operating principle that transfers to your own stack.
Skipping the artifact, which means the learning never becomes operational or inspectable.
Do not count this as learned until these are true.
01
State the transcript-backed claim in your own words: Mads walks through the open-source Claude Code job-search repo he built while unemployed after a December layoff: a profile builder, a multi-site job scraper, a scored fit evaluation, and a drafter/reviewer agent pair that writes grounded CVs and cover letters. It doubled as GitHub's number one repo of the week twice, and the interview doubles as an argument that packaging your workflow as shareable skills now beats writing a blog post about it.
02
Explain the practical stakes without hype: New playlist item from The Next New Thing; queued for transcript-backed review, topic mapping, and a practical learning artifact.
03
Map the idea onto the Use Case -> Workflow -> Agent Role -> Metric -> Risk -> Adoption sequence and name the weakest link.
04
Produce the artifact and include the evidence that proves it: A one-page business case for one agent workflow.
Put it into practice
Give this grounded prompt to Codex or Claude after watching.
You are helping me turn one specific YouTube video into real, durable learning.
Source video:
- Title: He Built an AI Job Search Agent After Getting Laid Off — Now It Has 30K GitHub Stars
- URL: https://www.youtube.com/watch?v=HoVxjMNFYv4
- Topic: AI Strategy
- My current learning frame: Build a two-agent version of a document you write often: one agent drafts from a ground-truth profile file, a second agent in a fresh context window audits every claim against that file, and package the whole loop as a reusable skill you could hand to someone else.
- Why this matters: New playlist item from The Next New Thing; queued for transcript-backed review, topic mapping, and a practical learning artifact.
Transcript anchors from this exact video:
- 0:30 / Evidence 1: "of the week twice. Now, we're going to do a screen share where he'll show you how this system automates a job hunt and how you can use it, too. But, the other takeaway that I want you..."
- 2:37 / Evidence 2: ">> And the way that you got it by was by running a prompt in Claude. You want to show that and show the output? >> Yeah. Sure. So, the prompt that I just wrote was this job..."
- 6:38 / Evidence 3: "letter and your CV for for this specific role. And then it's sends that off to a reviewer agent which has a fresh context window, so no no information of who you are and it's it's not really..."
- 9:39 / Evidence 4: "So I like the idea of still having some kind of human in the loop just at at some point in in in the workflow, right? And that's also why I came up with this evaluation fit, right?"
- 12:24 / Evidence 5: "I've been so happy about the open source community myself and all the things that that people put out there on on for example GitHub with the repositories and how they they share the knowledge. Um so, I..."
- 14:38 / Evidence 6: "That's the vision, right? >> It's also funny if you actually look at the Now, I don't have the statistics about how much of my repositories code and how much of it is actually just plain text. I..."
- 17:16 / Evidence 7: "um RAG systems and and different stuff uh uh you you use LLM as a judge. So, you you can use another agent to um critique or score one agent's output. So, using this basic idea was something..."
Your task:
1. Use the transcript anchors above as the primary source packet. If you add outside context, label it clearly as outside context and keep it secondary.
2. Create a source-check table with columns: timestamp, claim, what the demo proves, confidence, and what still needs verification.
3. Extract the actual teachable claims from the video. Do not invent claims that are not supported by the title, lesson frame, or transcript anchors.
4. Build a reusable learning artifact: A one-page business case for one agent workflow.
5. Include:
- a plain-English definition of the core idea
- a diagram or structured model using this sequence: Use Case -> Workflow -> Agent Role -> Metric -> Risk -> Adoption
- 3 concrete examples that apply the video idea to real agentic work
- 2 failure modes the video helps prevent
- a checklist I can use the next time I run Codex or Claude
- one practical exercise with a clear done signal
6. Add a "learning transfer" section: what changes in my workflow tomorrow if I actually learned this?
7. Add a "source check" section that cites which transcript anchor supports each major takeaway.
Quality bar:
- Make this specific to "He Built an AI Job Search Agent After Getting Laid Off — Now It Has 30K GitHub Stars", not a generic AI Strategy essay.
- Prefer operational examples, failure modes, and reusable artifacts over broad definitions.
- Call out uncertainty instead of smoothing over weak evidence.
- If evidence is weak, say what transcript segment or timestamp needs review instead of guessing.
- Finish with a concise artifact I could paste into my learning app.
Misconceptions
What to stop believing.
Every new AI tool deserves a trial.
Every tool has integration cost. Start from workflow pain, not novelty.
If an agent can do it once, it is automated.
Automation means repeatable, monitored, recoverable, and reviewable.
Practice studio
Learning only counts when you make something.
01
Transcript evidence map
Separate what the video actually says from what you already believe about the topic.
3 source-backed takeaways with timestamps, confidence, and a transfer note.02
One useful artifact
Apply the video to a real workflow and produce a one-page business case for one agent workflow..
A reusable artifact with a done signal and one verification step.03
Teach-back card
Explain the lesson to someone who has not watched the video yet.
A 90-second explanation, one diagram, one example, and one misconception to avoid.
Recall check
Answer first, then reveal — without rewatching.
What does the fit evaluation score a role on, and what is the score actually for?
Why does the reviewer agent run in a fresh context window?
Why did Mads deliberately leave auto-apply out of the repo?
Source shelf
Use the video as a doorway, then verify with primary sources.
