Broader evidence for the opening
Geoffrey asked for evidence beyond Holcim’s individual target: broad trends, what leading adopters do and credible executive quotations. The current presentation remains html-v10 with header branding. This record proposes a replacement, not an implemented slide change.
Editorial recommendation
Replace Holcim as the main opening exhibit with “AI leaders aim beyond efficiency.” Use the workflow-redesign comparison in B01 as supporting evidence, label its values approximately, and explain that the same research associates strong outcomes with broader growth and innovation ambitions. Keep Holcim as an optional spoken sector example or backup.
Then use the short B05 executive quotation as the bridge into engineering. The presenter’s own transition could be: “Our part of that challenge is making the valuable application practical: meeting its requirements for quality, response time, capacity, cost and control.” This is our interpretation of the sources, not a quotation or a claim that the surveys isolate engineering’s causal contribution.
Use one evidence exhibit in the opening, not a series of consultancy slides. Preserve the two demonstrations and the 23½-minute target. If a second quotation is wanted, B06 belongs beside control and ownership later in the talk. These sources support the strategic direction; our measured example and eventual verified demos must carry the technical proof.
Ranked options
| Priority | Source | Why it earns a place | Suggested treatment |
|---|---|---|---|
| 1 | B01: McKinsey, August 2026 | Strongest direct fit to growth, innovation and redesigned applications | Two horizontal bars with explicit approximate labels; use qualitative explanation in notes |
| 2 | B05: Nadella, December 2025 | Clear bridge from model capability to engineering a useful system | One short quotation; no separate evidence claim |
| 3 | B03: BCG, September 2026 | Connects leaders’ execution to focused programmes and measurement | Optional 61 versus 17 programme comparison; avoid implying a causal effect |
| 4 | B02: PwC, January 2026 | A broad financial-result gap | 56% neither / 12% both, with the question and period; keep the remaining categories clear |
| 5 | B06: Nadella, July 2026 | Unusually relevant to the control and independence discussion | Brief attributed excerpt beside control; avoid blanket claims about provider data use |
| Reserve | B04: BCG AI Radar | Spending intentions in financial units | Paired 2025/2026 bars; show planned status and the smaller industrial-sector increase |
| Reserve | B07: Siemens CES | Industrial leadership relevance | More rhetorical; less precise than the systems quotation |
Evidence discipline
The source register [local reference omitted] contains exact short quotations, dates, populations, locators and limitations. The chart data separates plans, reported outcomes and approximate narrative values. No chart has been inserted into the deck.
These surveys do not share one population, question, period or definition of value. Do not turn them into a funnel, pooled estimate or universal causal model. Selecting companies by high reported performance can itself produce associations with maturity and investment. A reported lack of revenue/cost gains does not establish failure. Likewise, prospective investment is not audited expenditure.
Avoid the older 2025 BCG “5% versus 60%” opening when a September 2026 successor is available. Avoid stitching different BCG studies into a time series. The broad-sample and industrial-sector budget movements differ, so the all-industry growth figure is not a Knauf-sector forecast. Keep CEO statements visibly separate from empirical evidence.
The McKinsey article’s workflow comparison is approximate; its original SVG endpoint did not load through the web tool and the direct fetch timed out. Use approximate labels, or retrieve the chart before publishing exact percentages. No exact chart labels or confidence intervals were invented.
Continuation
Research completed against private main 9e2453e; remote and open-PR checks showed no overlapping work. Next editorial step is selecting the opening exhibit and quote before revising the canonical scenes and narration. No website publication, slide changes, new demo claims or external messages in this research pass.
Integration into html-v11, 27 October 2026
Geoffrey requested integration of the numbers and quotations into the argument and flow. B01 now opens the evidence on slide 4 with approximate, proportional workflow-redesign bars. B05 provides a short bridge to the system constraints on slide 6. B06 supports the ownership discussion on slide 14. B03’s 95% measurement finding enters slide 16 narration, with a visible citation. B02 is backup A8; Holcim moves to A7. B04 and B07 remain research reserves.
The canonical script, curated audience summaries, source register and references are updated together. Main content remains 23:30, Klaus starts at 05:00, and speaker allocation remains deferred. Survey association is not causal proof; quotes are perspectives, not measured results. See the current revision record [local reference omitted] for implementation and verification.