YubHub
Job feed aggregation that actually works
Scrapes employer careers pages directly. Enriches every listing with AI. Delivers clean structured XML feeds to your job board. No bulk feed noise.
The problem
Bulk feed providers
- 50,000 random listings
- Duplicates everywhere
- Stale postings, dead links
- Missing salary, location, skills
- No vertical focus
YubHub
- You pick the employers
- Direct from careers pages
- Auto-refreshing, always current
- AI-enriched: salary, skills, level
- Built for niche verticals
I ran a sim racing job board for a while (don’t ask) and the data quality from the big aggregators was terrible. Duplicates everywhere, stale postings, irrelevant roles, and half the listings were missing basic fields like salary or location. So YubHub takes the opposite approach: you pick the employers, you build feeds around specific companies or industries, and because we go directly to the source rather than scraping other aggregators, you end up with jobs that literally nobody else has in their feeds.
How it works
Paste a careers page URL
Any employer, any ATS. Greenhouse, Lever, Workday, Workable, SmartRecruiters, Ashby, Oracle HCM, SAP SuccessFactors, Pinpoint, Phenom, SmartDreamers, or custom career pages.
AI enriches every listing
Each job gets salary ranges, skills extraction, experience levels, work arrangements, industry classification, and company logos. 100% coverage on skills, 99%+ on most fields.
Get a structured XML feed
Standard RSS 2.0 that plugs into jboard, Niceboard, Jobboardly, SmartJobBoard, WordPress, or anything else. Feeds refresh automatically on your schedule.
What you get per job
*Lower because many UK employers still don’t publish salary data. But 49.4% is high for this space, because our AI extraction catches salary information embedded in job descriptions that structured scraping misses.
Currently powering
SimRacingCockpit.gg
FluidJobs
Under the hood
Built on Cloudflare’s edge platform: six Workers coordinating through four Queues, D1 for the database, Workers AI (Llama) for enrichment. Firecrawl handles JavaScript-rendered scraping, Brave Search pulls company metadata, and a logo resolution service finds favicons automatically. No traditional servers. I’ve had it processing thousands of jobs a week for months without touching it.
For developers: full REST API, and an MCP server (@houtini/yubhub on npm) that lets you manage feeds through Claude Desktop. Which, frankly, is how I manage most of my own feeds at this point because it’s faster than the dashboard.
Built for
Niche job board operators
Employer-direct listings that match your vertical. If you’re running a board in a specific industry, the generic feeds are mostly useless and you already know it.
Recruitment agencies
Structured job data from specific target employers for market intelligence and candidate matching.
WordPress site owners
WordPress plugin (early access) that imports feeds as custom post types with Schema.org markup, search filters, and four prebuilt themes.
Programmatic platforms
Standard RSS 2.0 with all enriched fields. Plug it into your distribution pipeline.
Our first microsaas
YubHub is the first product shipped under the Houtini umbrella, and it’s a good example of how we approach software. The entire platform was designed, built, and shipped by one person using the same AI-augmented development workflow we teach in our consulting practice. Claude Code for the architecture and complex logic, local models via Houtini LM for the routine tasks, and a bias toward shipping something that works over designing something perfect.