# Scirium Mockup Review - Blind Feedback Sprint Date: 2026-08-16 Reviewers: 3 (blind, order-rotated, browser-rendered + computed-style verified) Moderation: all 3 passed anti-slop (specific, location-cited, blank-page test met) ## Verdict No clean winner. Cutting-edge and Premium tie at 7 points each; Corporate loses at 4. The fork is philosophical: product-proof (Cutting-edge) vs identity/memorability (Premium). Corporate fails on memorability, which is fatal in a crowded AI-chatbot market. ## Positioning check (target-user guess) Result: BULLSEYE. All three reviewers independently identified the product as a self-hosted, white-label, cited-answer internal knowledge assistant for enterprises on M365, scoped one-department-per-channel. Two reviewers also caught the MSP white-label reseller angle ("A product of IT Pro Partner" + multi-tenant + white-label). Positioning and copy are NOT the problem. ## Rankings (3/2/1 scoring) | Rank | Direction | Points | Reviewer ranks | |---|---|---|---| | 1 | Cutting-edge | 7 | 1, 1, 3 | | 1 | Premium | 7 | 2, 2, 1 | | 3 | Corporate | 4 | 3, 3, 2 | ## Per-direction read ### Cutting-edge - Pros: interactive demo is the single best element across all pages (shows the citation mechanic, not describes it); scannable comparison tags; "Billing answers billing. HR answers HR." best articulates the core invariant. - Cons: primary CTA white-on-transparent over a light page = near-invisible (verified); demo tabs inert beyond HR; 8-pill buzzword dump; reads "dev-tool startup" which is at odds with "self-hosted enterprise-grade." The "cutting-edge" label does not match the flat light render. ### Premium - Pros: only genuinely memorable identity (Fraunces serif + cream + teal); best copy ("Trust is the feature. The cited source is the product."); strongest citation proof (full inline [1]/[2] footnotes). - Cons: hero nearly empty with no product visual; "See it answer →" is ungrammatical and promises a demo never shown; lazy-render blank scroll gap (verified via computed styles, reads as half-broken); "One undifferentiated model is not an answer." is jargony and buries its own good table; buzzword pill strip. ### Corporate - Pros: clearest 10-second read; highest enterprise credibility; best single trust line ("It says I don't know instead of guessing."); trust badges with icons. - Cons: unanimously "generic / forgettable / stock Tailwind template"; ASCII "+" prefix in comparison table (2 reviewers); inconsistent category abbreviations (BI/ON); leads with static image instead of a demo. ## Blocking fixes (consensus, ranked by recurrence) 1. Premium: empty hero, no demo - add the interactive cited-answer demo (2 reviewers). 2. Cutting-edge: CTA legibility - white on transparent over light bg (verified, 1 strong). 3. Both: 8-tag buzzword pill strip is filler - cut or reduce to 3 hard facts (2 reviewers). 4. Corporate: "+" ASCII table clutter + generic identity (3 reviewers unanimous on generic). 5. Cross-cutting: scroll-reveal produces a blank gap that reads as broken content (flagged on multiple pages, verified). ## Synthesis recommendation None of the three perfectly lands "premium enterprise AI" - Cutting-edge overshoots toward startup/dev-tool, Premium overshoots toward editorial/law-firm, Corporate undershoots into generic. The end state every reviewer converged on is the same: - Cutting-edge's interactive demo (the proof) - Premium's identity + copy (the memorability) - Corporate's trust-badge row + "says I don't know" line (the enterprise signal) Recommendation: converge the winner on Premium's identity/copy as the brand, rebuilt around Cutting-edge's demo, with Corporate's trust elements ported in. Tie-break rationale: for an enterprise IT/ops buyer, memorability + trust beats startup energy; the demo is the asset to carry over, not the aesthetic.