KiTTY VS getAIwork

getAIwork's answer:

The published articles run on WordPress.

The Work Board and the pipeline behind it are custom built. Listings are ingested from multiple job sources several times a day, passed through an automated screening and de-duplication pass, then queued for human approval before anything appears publicly. Filled and expired roles are removed on the same daily cycle.

Aggregate statistics are generated from that pipeline and published as static JSON at getaiwork.com/stats.json under an open attribution licence, alongside a human-readable version at getaiwork.com/stats.

getAIwork's answer:

It started from screening job postings in bulk and noticing how much of what circulates about AI work is either out of date or simply invented. Boards were full of roles that had closed weeks earlier, and articles quoted earnings figures nobody could source.

The response was to build the boring version. Screen everything, have a human approve what goes live, delete listings the day they are filled, quote pay only as the posting states it, and publish the underlying counts openly so anyone can check them.

The training and certification grew out of the same work. Screening tens of thousands of postings shows you fairly clearly what employers actually ask for, which turns out to be a better curriculum than guessing.

getAIwork's answer:

Only if the screening is what you care about. If you want maximum volume, a large aggregator will always list more roles. getAIwork deliberately lists fewer, because a person removes the ones that are filled, duplicated or not genuine.

The Work Board is also free and public, with no signup and nothing to buy. That is unusual in this category, where screening is normally the thing kept behind the paywall. The paid membership is the training and the human-graded certification, not access to the listings.

And pay is shown exactly as each posting states it, never estimated or modelled. There are no income claims anywhere on the site, which rules getAIwork out for anyone looking to be told what they will earn.

getAIwork's answer:

Two things, both fairly unglamorous.

First, every listing on the Work Board is approved by a person before it goes live, and listings are deleted the day the role is filled rather than left to decay. Most boards in this space aggregate automatically and let expired postings pile up, which is the standing complaint readers have about nearly all of them. Removing listings is the unglamorous, permanent job that makes the board worth trusting.

Second, the underlying market data is published openly instead of being kept as a marketing asset. 39,389 postings screened to date, roughly 1,363 live at any one time, and listed pay aggregated into percentile bands by field. It is published at getaiwork.com/stats and as machine-readable JSON at getaiwork.com/stats.json, free to reuse with attribution, so everything claimed above is checkable rather than asserted.

getAIwork's answer:

People trying to get into AI work, rather than people already established in it. Roughly three groups.

Career changers with no AI background looking for a realistic entry point. Freelancers and contractors adding AI work to what they already do. And people looking specifically for recurring paid platform programmes, meaning annotation, evaluation and model-training work, which is now close to half of everything open on the board.

The audience is global rather than US-centric, and a meaningful share are non-coders: writing-centred work is about 27% of currently open listings, against 68% technical.

On remote, the honest figure is that 564 of 1,351 currently open listings are tagged remote, about 42%. That share has drifted down slightly since late August rather than up, so this is not the remote-only market it is often described as.