Live trend ingestion
Every generation fetches fresh items in parallel from Hacker News, 11 finance and technology subreddits, 10 industry RSS feeds, and News API. Nothing is cached between clicks, so posts reflect what was published today.
netpost
LinkedIn Post Generator
Netpost writes five LinkedIn posts per click for B2B fintech and investment banking teams. Each one is grounded in a live industry trend pulled seconds earlier, checked against a linter for AI-sounding filler, and linked back to its source. No login, no account, free to use.
What it is
A trend-aware writing pipeline, not a prompt box
Most AI writing tools start from a blank prompt and produce confident, sourceless copy. Netpost starts from what the industry actually published today. It reads live trends, ranks them, and writes only about the ones that scored — then shows you the score, the flags and the links so you can check it.
- Items ingested per run
- ~75
- Live sources
- 4
- Posts per click
- 5
- Login required
- None
The pipeline
Four stages, run fresh on every click
Ingest
Roughly 75 items per run are fetched in parallel from Hacker News, 11 subreddits, 10 industry RSS feeds and News API. No API keys are needed for the first three.
Dedupe and score
Normalized titles are hashed to SHA-1 fingerprints to drop near-duplicates. What survives is scored on keyword relevance × source weight, plus a freshness bonus of up to 2.0 for items under 24 hours old.
Brief
The top 5–10 items become the trend brief, each tagged (Fintech, Automation, Banking) and carrying a plain-language reason it was ranked where it was.
Draft and check
Five posts are generated, each mapped to a different trend, then run through the linter and the similarity checker. Anything duplicated or too close to prior work is regenerated, up to three attempts.
Where the signal comes from
Five live inputs, weighted by reliability
Sources are weighted before ranking, so a trade-publication story outranks a forum thread on the same topic. Company blogs and changelogs are supported through configurable RSS entries rather than a separate connector.
| Source | Per run | Weight |
|---|---|---|
| Hacker News | Top 20 stories | 1.1 |
| 4 posts × 11 subreddits | 0.95 | |
| RSS | 6 entries × 10 industry feeds | 1.2 |
| Company blogs | Changelogs and releases via RSS | 1.3 |
| News API | Up to 20 articles | 1.4 |
What you get
Every post arrives with its evidence attached
Dedupe and relevance scoring
Items are fingerprinted with SHA-1 hashes of normalized titles to drop near-duplicates, then scored on keyword relevance multiplied by source weight, plus a freshness bonus for anything under 24 hours old. The top 5–10 items become the trend brief.
Researched style patterns
A curated corpus of high-performing B2B creator patterns supplies the hook types, structures, credibility moves and pacing that actually perform on LinkedIn. The corpus stores pattern summaries only — never copied posts.
Two brand voices
Founder voice is sharper, more opinionated and operator-led. Company voice is measured, educational and category-authoritative. Both draw from the same trend brief.
Anti-slop linting
Each post is scored out of 100 and flagged for weak hooks, vague claims, filler CTAs, hashtag spam, missing credibility and unreadable paragraph length — the tells that make writing read as machine-generated.
Similarity checking
Every draft is compared against the style corpus and your recent posts using token overlap, 3-gram shingles and cosine similarity. Anything scoring 45 or above is blocked before you ever see it as publishable.
Source grounding
Each post carries the trend it came from and links back to the original articles, so any claim can be checked against its source before you publish.
Performance feedback loop
Record impressions, reactions, comments, reposts, saves and clicks on posts you publish. Netpost summarizes what worked by hook type, format and voice, and feeds that into the next batch.
How it is built
Tech stack
| Layer | Stack | Host |
|---|---|---|
| Frontend | Next.js 15 App Router, React 18, TypeScript, three.js | Vercel |
| Backend | FastAPI, Python, httpx, feedparser | Render |
| Database | Supabase (PostgreSQL) via REST | Supabase |
| Generation | OpenAI chat completions with structured JSON output | OpenAI |
A deeper breakdown of the scoring maths, the linter rules and the similarity thresholds lives in how Netpost works.
Questions
Frequently asked questions
What is Netpost?
Netpost is a free LinkedIn post generator built for B2B fintech and investment banking teams. It pulls live industry trends, layers in researched style patterns from high-performing B2B creators, and returns five ready-to-edit posts per click.
How does Netpost work?
Netpost runs a four-stage pipeline. It ingests items in parallel from Hacker News, Reddit, RSS feeds and News API; deduplicates them with SHA-1 title fingerprints and ranks them by keyword relevance, source weight and freshness; builds a trend brief of 5–10 items; then generates five posts with OpenAI, each mapped to a different trend and checked by a linter and a similarity checker before it is returned.
Is Netpost free, and do I need an account?
Yes, Netpost is free and there is no login, no account and no signup. Open the site, pick a voice, and press Generate Posts.
Where do the trends come from?
Four public sources: the Hacker News top stories, 11 finance and technology subreddits, 10 industry RSS feeds including Finextra, PYMNTS, American Banker, Banking Dive and Finovate, and News API. Roughly 75 items are fetched per run and ranked down to the strongest 5–10.
How does Netpost avoid generic AI writing?
Three mechanisms. A linter scores each post out of 100 and flags weak hooks, vague claims, filler CTAs and missing credibility. A similarity checker blocks drafts that are too close to the style corpus or your recent posts. And every batch carries a unique nonce, fresh angle targets and a list of recent hooks to avoid, so repeat clicks do not converge on the same phrasing.
What is the difference between founder voice and company voice?
Founder voice writes as an operator: sharper, more opinionated, first-person and willing to take a position. Company voice writes as the brand: measured, educational and category-authoritative. Both are grounded in the same live trend brief.
Generate your next five posts
Pick a voice, press the button, and read the trend brief behind every draft.
Back to the generator