I stopped hiring for portfolios and started hiring for explanations
The best portfolio in the pile belonged to the person I did not hire.
LinkedIn posts from the work you already do
You already did
something worth saying.
ThreadsMark turns the work you are already doing into LinkedIn posts written in your voice. It keeps what you capture, finds what is worth saying, drafts it from your own detail, and posts it on the schedule you set. Nothing is invented, and nothing goes out until you approve it.
- Consultants
- Recruiters
- Founders
- Designers
- Marketers
- Finance
- Engineers
- Agency owners
Official LinkedIn API · never asks for your password
Tuesday, 4:12pm · your vaultraw note
Rejected the candidate everyone else loved. Beautiful portfolio, but they couldn't explain a single decision in it. Hired the one who walked me through their worst project instead. Best hire I made all year.
Your interview process is measuring the wrong thing
Everyone screens for what someone made. Almost nobody screens for whether they know why.
The best hire I made talked me through their worst project
I asked what went wrong. They had the answer ready — and that was the answer.
One note. Three arguments, three readers. Each one traces back to what you wrote.
Material comes to you
Connect a blog, GitHub, Notion or Slack, or drop in documents and meeting notes. What you publish, merge and write arrives as something to keep or skip. Nothing is published on its own.
- Any blog or newsletterLive
- GitHubLive
- NotionLive
- SlackLive
- Documents and pasted notes
Every week
You did 4 things worth talking about this week.
You will not remember them by Friday. ThreadsMark does, because it read them on Tuesday when you wrote two lines about the thing that went wrong. Your week gathers them, with the angles each one could become.
Every angle points back at the note it came from. Nothing is invented, and anything it cannot source, it drops.
Your week · an example4 captured · 6 stories
Rejected the candidate everyone else liked. Hired the one who walked me through their worst project.
What a portfolio cannot tell you about how someone thinks
The interview question that changed who I hire
Spent two days on a bug that turned out to be an eight-month-old doc nobody had read.
The onboarding doc nobody reads is a bug report
Killed the Monday status meeting. Three weeks in, nobody has asked for it back.
We deleted a meeting and nobody noticed
What a written update forces you to admit
Quoted six weeks, client wanted three. Held the line and explained what would get cut.
Pushback on a timeline is usually about confidence, not the number
Pick what you want to share. Nothing goes out until you approve it.
Technical breakdownShort insightPersonal storyContrarian opinionLesson learnedBuild in publicCase studyObservationCareer reflectionDeep explanation
The loopReal work in. Writing out.
Real work in. Writing out.
It gets sharper the longer you use it.
Step 01 of 5
Ten seconds, no structure.
Dump a rough thought and close the box. ThreadsMark reads it afterwards, works out what kind of thing it is, pulls out the situation, the decision and the lesson, and flags anything that looks confidential. Your words are never rewritten.
experiences · structured asynchronously
Before anything publishes
Your vault is not your feed.
Most of what you record should never be published. Mark something private and it is excluded from every prompt — not filtered afterwards, never sent at all. And before anything goes out, this runs.
Your draft — edit itcheck before posting
runs in your browser · nothing sent anywhere
2 findings
AWS access key idhigh
AKIAIO…MPLE
This looks like an AWS access key id.
Remove it and rotate the key — anything posted publicly should be treated as compromised.
Email addresslow
dana@e….com
This is an email address. If it belongs to someone else, they did not consent to it being posted.
Remove it unless it is your own and you mean to share it.
That is the real scanner — the same module that runs in the product, executing in your browser on text you control. It never blocks you from posting. You are the one who knows what your NDA says; its job is to make sure you noticed.
On authenticity
It will not invent a version of you.
What it will never do
- Invent a metric you never recorded
- Turn "faster" into "40% faster"
- Claim a project you did not work on
- Give you an opinion you never held
- Write about a subject you ruled out
What it does instead
- Cite the vault entry it drew from
- Say where it had to write around a gap
- Ask you the question it needs answered
- Leave the number out and write around it
- Drop the idea entirely if nothing backs it
This is not a guideline in a prompt. An idea that cannot cite a real vault entry is discarded before you ever see it, and a draft that had to work around a missing detail has to say so.
Why trust it with your name
The boring parts are the ones that matter.
0
blind retries
Each approved post is claimed once before it is sent, and a LinkedIn post can only be recorded against one of yours. If LinkedIn’s answer is unclear, the post is held for you to check instead of being sent again.
0
checks before posting
A pattern scan for keys, tokens and personal data, and a read for what a pattern cannot catch, like a client you should not name. Neither one blocks you.
0
named editing actions
Each changes the one thing you asked for, and saves a version first. No prompt box that can quietly rewrite everything.
0
LinkedIn passwords asked for
Posting goes through LinkedIn’s official API and sign-in. Nothing logs in as you or drives a browser on your behalf.
You have been meaning to write more for about two years now.
Connect LinkedIn, write down one thing that happened this week, and ThreadsMark takes it from there: drafted in your voice, approved by you, posted on your schedule.