ALPHA Timepoint is in alpha Talk to Us
← Blog · July 17, 2026 · Sean McDonald

Picking next year's flagship format? Run the slate before you greenlight it.

If you run programming for a media brand, you know the quarter this post is about: three credible format bets, one production slot, and a greenlight meeting where the loudest conviction usually wins. The research that would settle it — a pilot, a panel, a soft launch — costs months you don't have. There's now a way to run the same question in days, as a simulation. Here's what that looks like.

New here?

Timepoint AI is a Santa Monica company that runs decisions as grounded simulations: the specific, named actors around your decision, played forward through branching timelines. Findings come back ranked, with uncertainty stated in words — never sold as a prediction, always labeled as the AI simulation they are. Here's the full range of work →

A slate decision is a systems question

It feels like a taste question, but it isn't. The audience you'd win, the audience you'd cannibalize, what the rival programmer does the week you announce, how the platform's discovery machinery treats each format — every one of those is a moving part with its own incentives. A deck argues about them. A simulation runs them: synthetic audience panels for each format, the competitor's slate meeting played from their side of the table, and the branch where your bet underdelivers, so you know what breaks first.

SIMULATED SCENARIO — FICTIONAL DEMONSTRATION. "The studio," its formats, and its rival and every finding below are invented to show how the method runs (FTC 16 CFR 465). This is an AI simulation of a fictional situation — not market research, not a real engagement, not client traction.

A culture-and-competition media house (fictional) has one flagship slot for 2027 and three candidates: a prestige docuseries following its athletes, a live head-to-head competition format, and a creator-led short-form series with weekly drops. The intake call took nine minutes. Here's the shape of what came back:

THE DECISION The 2027 flagship slot 1 Creator-led weekly MOST STABLE ACROSS RUNS 2 Prestige docuseries SAFE, FLAT, DURABLE 3 Live competition HIGH DIVERGENCE — EXECUTION BET RANKED BRANCHES · UNCERTAINTY STATED IN WORDS · AI SIMULATION, LABELED
The deliverable's skeleton: one decision, three branches, ranked — with the stability of each ranking stated, not hidden. Dashed = the runs disagreed; that disagreement is itself a finding.
RankFormatWhat the runs showedStated uncertainty
1Creator-led weeklyThe only format whose simulated audience was additive — new viewers, not reshuffled ones. Rival counter-programming barely touched it: their slate meeting (simulated) had nothing shaped like a response.Most stable across runs. Sensitive to one thing: two named creators actually renewing.
2Prestige docuseriesSafe, flat, durable. Every run delivered the core audience and nothing else. The stress branch degrades gently — a miss costs prestige, not reach. Low divergence; runs agreed on direction and size.
3Live competitionThe biggest ceiling and the widest spread. It's the format the rival's simulated programmers wanted picked — the one their existing rights and talent could counter inside a quarter.High divergence — a bet on execution and timing, and the runs say so plainly.
The blind spot the backward run surfaced

Every forward argument in the (fictional) building compared the formats on quality. Reasoning backward from the 2028 outcomes — Timepoint calls this portal analysis — showed the real divider was discoverability: the platform's recommendation machinery treats a weekly-drop creator format and a live event as different species, and that structural difference dominated the quality difference in most timelines. Nobody had put it on a slide, because it wasn't anyone's job to.

What your team would walk away with

A ranked brief the greenlight meeting can argue with instead of about: which audiences each format actually moves, the rival's most likely counter, the assumption each ranking leans on, and the early signal that would tell you — weeks in, not seasons in — that the bet is wrong. It's scoped in one fifteen-minute call, and when the facts change, the same simulated world re-runs without starting over.

See it in practice

Read this honestly The worked example above is fictional and labeled, because we don't publish client work — and we don't publish accuracy percentages, ours or anyone's, because no published calibration record substantiates one yet (here's what we test instead). What real engagements share with the fiction is the form: ranked branches, uncertainty stated in words, disagreement between runs reported instead of averaged away, and every output labeled the AI simulation it is.
Bring us a decision What we do Pricing & engagement