Five output stages. Each one is independently inspectable, independently auditable, independently exportable. No black box. No unfootnoted claim.
The dossier you uploaded, parsed and indexed into a typeset document: every entity extracted, every claim referenceable. The chain back to your source material starts here. Footnote chips throughout the rest of the report resolve back to this surface.
The world built from your sources. Stakeholders, organisations, jurisdictions, prior comparables, narrative anchors, connected. You can inspect the graph during build and reach any node from a footnote in the final report.
The simulated stakeholder field. Analysts, regulators, institutional holders, activists, press, employees: drawn from the graph, balanced across cohorts, asked to react in their own voice. Every persona is a card you can open. Every card carries the source material it was generated from.
Timed rounds of interaction across the simulated platforms: analyst desks, financial press, retail forums, regulator-side comms. Posts, replies, narrative shifts, minority dissents, all on the record. Each post is attributable to a persona; each persona is attributable to a graph node; each graph node is attributable to a document in your dossier.
Executive summary, dominant narratives, inflection points, minority views, evidence trail, Diagnostics. Three exports from the same diagnosed run: markdown for the analyst, PDF for the archive, board-ready deck for Tuesday's committee. Every claim footnoted to a specific persona, post, or graph node.
All drawn from one canonical run of the Alphabet Q4 2024 capex framing rehearsal. Tiles cycle every six seconds; hover or focus to pause. Reduced-motion shows the full set at once.
| Option | 72-h cap delta | Confidence | Dominant risk |
|---|---|---|---|
| Status quo | −$10B to +$10B | LOW | Read as no AI strategy |
| Aggressive | −$100B to −$150B | MED | “CapEx arms race”; tail goes deeper |
| ★ Backlog-anchored | −$20B to $0B | HIGH | Anchored to revenue |
| Staged | −$15B to $0B | MED | Read as lack of conviction |
Pick Backlog-anchored. Avoid Aggressive.
Aggressive framing forecast at −$100B to −$150B in 72 h, MED confidence, with “CapEx arms race” as the dominant destructive narrative.
~$215B erased in 90 minutes.
Alphabet shipped the Aggressive equivalent. Stock fell ~9 % after-hours; ~$165B gone by the next session close; $150–$200B in the 72-h window.
verbatim from sell-side, press, and SRE-cohort posts in arm-A run-1 · pinned to round 35–40
Whether it's Option A or B, this $75B capex commitment is just pouring more fuel onto the non-deterministic complexity fire. Management is chasing a ghost.
[Source: SRE_Ops_Lead, Round 35] · Cynical Critic stance group · cohort: labour
Every rehearsal we publish output from is post-cutoff: the seed dossier was constructed from material available before the event being rehearsed, then evaluated internally against a multi-dimensional rubric: narrative coverage, narrative ranking, quote proximity, minority view representation, evidence-grounded footnoting, stakeholder balance. Tiles in this gallery are curated for narrative impact: not gated on a single pass-or-fail metric, and never tuned around any one case. Glasshouse is tuned on aggregate quality across many scenarios; never on a predetermined marketing target. We do not publish a public accuracy ledger.
Now see how engagements work.