Built on 31 peer-reviewed findings
Slopster is an autonomous content engine. Give it a brand and a goal. It generates a population of candidates, ranks them with a bias-controlled judge panel built on published research, refines the winners, packages them for posting, and learns from real performance.
You are pricing this wrong.
Nobody tells you this about churn.
I ran 40 tests. One worked.
Stop posting daily.
Why does this ad feel wrong?
Three tools, zero budget.
Wait for it.
The 4pm slump is a lie.
Here is what we shipped.
POV: your first customer.
Do this before Friday.
How we cut CAC in half.
Evidence
We did not guess at what makes content work or how to judge it. The engine is an implementation of the literature on short-form performance, LLM judging and online learning.
Read the full paper listCuriosity and pain-point hooks raise engagement over direct informational openings, holding video length, subtitles and face visibility constant.
In Slopster: Hook style is a first-class attribute. Generators must cover the taxonomy; judges score hook arousal explicitly.
Strong judges reach about 80% agreement with humans but show position, verbosity and self-enhancement bias. Self-preference is driven by perplexity: models rate familiar text higher.
In Slopster: Judges never share a family with the author they score. Every pairing is judged in both positions and disagreements are discounted.
Heterogeneous role-specialized agents with learned consensus beat homogeneous debate by 4 to 6 points and cut factual errors by over 30%.
In Slopster: Generation uses role-specialized authors (hook-smith, storyteller, direct-response, contrarian) across model families.
Generative AI improves individual creativity but collapses collective diversity.
In Slopster: Embedding-distance constraints at the angle and population stages; near-duplicates are culled before judging.
Batched Thompson sampling beats test-then-rollout for headlines by 3.69% clicks. Batch-updated bandits handle delayed rewards and delivered +6% CTR and +16% CVR in production.
In Slopster: Each brand keeps Beta priors per attribute value, updated in batches from real metrics and sampled during generation.
Humor and emotion raise engagement; informative content lowers it alone but raises it when combined with personality. Narrative content has the strongest effect on purchase intent.
In Slopster: The judge rubric scores a personality × information blend and narrative pull. Sales briefs are rewritten to carry both.
How a run works
A run is a search. It widens on purpose, narrows with a tournament, then hardens the leaders through debate before it packages anything. Every stage retries on failure and streams progress live.
Loads the brand, campaign, brief, learned priors, top past winners by embedding similarity and live trend signals.
A strategist proposes distinct angles. Angles are embedded and any pair that sits too close is regenerated.
Role-specialized authors across model families write structured candidates: hook, beats, shot list, on-screen text, CTA, caption.
A Swiss-style pairwise tournament. Judges come from a different family than the author and see both orderings.
A critic attacks the leaders, the author rewrites, a judge decides if the rewrite wins. Winners re-enter the bracket.
Finalists get imagery, voiceover, optional video, captions and hashtags tuned to the target platform.
Real metrics update per-brand priors with batched Thompson sampling. Judge weights are calibrated against realized rank.
What you get
Hook, timed beats, voiceover, on-screen text and framing for every second of a vertical video.
Scroll-stopping first lines, slide decks with image prompts, captions and platform-fit hashtags.
Bradley-Terry rankings over hundreds of pairwise verdicts. Absolute scores explain, they never decide.
Critics attack the leaders. Authors rewrite. A third-family judge decides who advances.
Published posts feed real metrics back into per-brand priors. Next run starts smarter.
Banned words, claims to avoid, must-includes. Every model call is metered against a hard cap.
Pricing
Every plan includes model usage. Runs are metered to the cent and stop at your budget cap.
Compare plansFor one brand finding its voice. Fast and standard runs, no video.
For teams shipping on a cadence. Max runs with video, scheduling and the learning loop.
For agencies and multi-brand operators. Volume pricing, SSO and a dedicated pipeline.
Paste a website. Slopster extracts the brand. Your first fast run costs about two cents.