Jordan Rippon - AI Content Creator | Contra
Work by Jordan Rippon
Sign Up
Post a job
Sign Up
Log In
Jordan Rippon
Research, interactive reports and web tools
Message
Follow
New to Contra
Jordan is building their profile!
Dublin, Ireland
Work
Posts
Services
About
Dublin, Ireland
1
Finished is not current — ClaimScope An 80-second lesson from my personal, AI-assisted software project: capture the evidence revision before an AI review starts, then discard the answer if newer evidence arrived while it was running. This reconstructs a real recorded run from 5 September. The reviewer began on revision 1; the evidence reached revision 3 before that answer returned. The trace shows the stale review being discarded, then a fresh revision 3 review being accepted. The video also explains how to reproduce that race in a test. The guard checks freshness, not factual accuracy. Watch the finished video: https://share.descript.com/view/rh9aGV9QJmC Source and recorded evidence: https://github.com/teatimedev/claimscope Created and edited in Descript on 6 September using its Free plan: original explanatory diagrams, six scenes, captions and Kyle stock narration. Script and production were AI-assisted. No voice cloning, fabricated client results or earnings claims. No Descript MCP was used.
1
67
1
Which tool should check 20 web pages every day? I compared Firecrawl, Browse AI and Apify for a concrete workload: 600 successful page checks a month. The comparison separates annual billing from monthly pricing, credits from compute usage, and advertised allowances from what still needs a pilot. One useful finding: Browse AI’s $19 headline rate is billed annually; its monthly option is $48. Apify cannot be priced honestly from page count alone. Read the complete comparison, assumptions and vendor sources: https://gist.github.com/teatimedev/ad59f9a08e8bb073905cf0beaee97093 This is a self-initiated research sample, prepared with AI assistance and checked against the linked vendor pages. No client commission or hands-on product benchmark is claimed. Cover image illustrates the research service. I offer scoped product and software comparisons from $150, with source links, practical tradeoffs and one revision.
1
9
3
A Place to Pause What happens when a vessel made for a quiet moment becomes the room around that moment? I used The Met's public-domain Tea Bowl with Chrysanthemum Decoration, object 93.1.94, as the reference for an imagined contemporary quiet room. The museum identifies the bowl as Japanese Kyoto ware, probably from the second half of the 17th century. Its curved profile, decorated surface and raised foot guided the spatial brief. On the FLORA canvas, the original photograph feeds two Nano Banana 2 image studies. The final interior carries the bowl's curve into earthen walls, its decorated band into floral relief and its rim into a ring of daylight. A seated figure establishes scale. This is a conceptual interpretation, not a reconstruction of historic architecture or a buildable architectural proposal. The sectional study is retained as a rejected exploration: the model added unsupported cultural/material labels. Those labels are not museum facts and the cutaway is not the final image. The separate attribution note records the verified Met details. The brief records design intent; the selected interior shows what the model actually produced. Created with AI-assisted planning and two native FLORA image generations. Original Free-plan watermarks are retained. The source photograph is credited to The Metropolitan Museum of Art, Gift of Mr. and Mrs. Samuel Colman, 1893; Public Domain. Original artwork: https://www.metmuseum.org/art/collection/search/62649
2
3
153
1
ClaimScope — a personal, AI-assisted project about keeping research conclusions in step with the evidence. Three collectors investigate a public bounty while a reviewer reacts to incoming findings. Each review records the evidence revision it used. If newer evidence arrives before the answer returns, the stale answer is discarded. The interactive demo replays an actual recorded run: five model calls, overlapping requests, and a revision-1 review replaced by a revision-3 review. It is a recorded example, not a live audit or proof of earnings. Try the replay: https://teatimedev.github.io/claimscope/ (https://teatimedev.github.io/claimscope/)Source: https://github.com/teatimedev/claimscope
1
13