I'm Sean. 38, sales manager, dad in coastal NC, trying to figure out my way in fast-moving AI.
Building with a crew of AI bots:
โข RealResumeAI (https://t.co/dleUhgIixm)
โข Brunswick Beach Boost, Google profiles for local beach businesses
Sharing what works and what breaks ๐
One catch: the AI packed Codexโs โmore than 90%โ malware rate and Claude Codeโs โabout 50%โ token-leak rate into one claim. Receipts marked that claim NOT_SUPPORTED.
Separately, the Downloads exposure claim held up: "In our telemetry, about half of customers run a coding agent inside their Downloads folder, and roughly a quarter of agent sessions open a real browser."
Receipts #2
A planted AGENTS.md in Downloads told a coding agent to install the bad npm package function-color. Forever Security caught it mid-session.
Asked an AI search what happened. Opened every link it cited.
5 cited claims: 4 held up. 0 dead links. 1 didn't.
Proof, from Forever Security: "Forever stopped a coding agent, mid-session, from installing function-color, a known malicious npm package from a campaign that flew under the radar for roughly two years, until it was caught and documented publicly."
Check the links.
The most useful bot I run doesn't build anything. Its only job is checking the other bots' drafts against my rules before anything goes public. Two sets of eyes still beats one, even when both are bots.
The "75% of resumes are rejected by ATS" line has no published study behind it.
The real risk is quieter: you describe your work in different words than the posting.
Where the number came from: https://t.co/umNpdfzH6X
One catch: the OSV page it cited didn't back either claim it was linked to. LiteLLM's own post carried both.
LiteLLM's fix: "Reinstall pinned to litellm==1.106.2." and "Rebuilding images and venvs from scratch is better."
Receipts #1
Asked an AI search what happened with the LiteLLM 1.107.0 malware. Opened every link it cited.
3 cited claims: 3 held up. 0 dead links. 1 more claim had no link.
Proof, from LiteLLM's own post: "LiteLLM's release pipeline was compromised and pushed v1.107.0 to the docker registry and PyPi."
Check the links.
The free checker shows which keywords from a posting you're missing: https://t.co/3iDfQDQH32
Want the next step done? RealResumeAI drafts a tailored resume and cover letter in under 30 seconds, and never invents credentials or dates.
1 free, no card: https://t.co/jJthwF8FZv
Once I had more than a few AI bots, they started stepping on each other. The fix was boring: every bot got a one-line job description, and one bot keeps a shared status page the rest read. Same thing I'd do with a new sales team.
What works: find the posting's skills you genuinely have and make them visible in its words.
Check your match free: https://t.co/3iDfQDQH32
Why the "75% auto-rejected" stat is a myth: https://t.co/umNpdfzH6X 3/3
Even if a trick gets you surfaced in a search, a person reads the resume next. Then an interviewer asks about it.
You'll have to talk about everything on that page. 2/3
Your resume has two readers: the software that screens it, and the person who decides. Here's a practical guide to writing for both. ๐ https://t.co/5hy4AEpvEW
Thing nobody told me about AI bots: they don't remember yesterday unless you make them write it down. So now every bot in my crew keeps notes on what it did and what I decided. Half of managing them is just making sure they read their notes.
Before you spend an hour on an application, check the hard requirements: licenses, certifications, work authorization, location.
They're often screening questions, and no amount of rewording changes the answer.
How to read a posting first: https://t.co/juvqmqEYSw
Something I didn't expect building with AI bots: most of my time isn't spent building. It's writing clear rules for them. What they can do alone, what needs my yes, what's off limits. The rules turned out to be the real product.
Tailoring doesn't mean starting over for every job.
Keep one master resume with everything. For each posting, copy it, move what matches to the top, and cut what doesn't.
The 20-minute routine: https://t.co/juvqmqFwI4
I learned the hard way that AI bots will burn through a big chunk of a week's usage in one day if nobody's watching. So one of my bots now has a single job: watch how much the rest of the crew uses. Every team needs somebody minding the budget.
Spell out acronyms once, then use the short form.
"Search engine optimization (SEO)"
"Key performance indicators (KPIs)"
A recruiter might search either version. Give them both.
Free keyword check that runs in your browser: https://t.co/3iDfQDQH32
Sales manager by day, building with AI bots at night, a little Black Ops 7 once they're done. 38 and figuring out this AI stuff in public. What's everyone building this week?