Audit Packet: Redesign the Work, Don't Bolt AI onto Old Jobs
This public Audit Packet documents the evidence basis, claim boundaries, counterarguments, editorial judgments, and falsification tests behind Brief No. 026.
This audit packet supports Brief №026: Redesign the Work, Don't Bolt AI onto Old Jobs. Read the brief first for the full argument.
Autonoma briefs are designed to be inspectable. This packet shows what the brief claims, how each claim was tested, what it does not claim, and where caveats remain — without exposing raw internal logs, prompts, source-routing mechanics, hashes, local paths, secrets, or unpublished candidate claims.
← Open Brief №026 — Redesign the Work, Don't Bolt AI onto Old Jobs
Brief Summary and Audit Verdict
Brief 026 asks whether the AI gains people report sit with redesigned work or with added tools, and what that means for L&D. Its closest neighbors are Brief 018, on what remains of an existing role, and Brief 025, on how new hires learn judgment.
Audit verdict: Supported at the association layer. A worker survey carries the redesign, training, and clarity gaps; an index, the return premium; interviews, the new roles; a Microsoft survey, the organizational share. An executive survey bounds any causal reading. BCG's "more than 11 percentage points" a year is rejected as overstated. No source measures output before and after a redesign. The join and the 12-month timing are Autonoma synthesis.
Claim Register
| # | Claim in the Brief | Status | Domain | Bound |
|---|---|---|---|---|
| B026-C01 | Reshape or Invent vs Deploy-only respondents: measurable business improvement 67% vs 43%, process redesign 43% vs 12%, major reskilling 65% vs 47%. | Supported | web-assets.bcg.com | June 2026. n=10,990. Self-classified. Correlational. |
| B026-C02 | 88% believe they need major upskilling; 36% feel properly trained, both unchanged from 2025. 66% of frontline regular AI users get limited or no guidance on time saved. | Supported | web-assets.bcg.com | n=11,749. Guidance base: 8,989 regular AI users. Self-report. |
| B026-C03 | Measurable impact: 55% with limited clarity and tools, 60% with strong tools only, 80% with strong clarity only, 83% with both. | Supported | web-assets.bcg.com | n=9,923 regular AI users. Self-reported. Strong tool access is a low bar. |
| B026-C04 | 47% agree their roles shifted toward managing and directing AI (30% disagree). Half of agent-aware AI users report no clear guidance on managing teams of people and AI. | Supported | web-assets.bcg.com | n=11,749 and n=8,849. Self-report. |
| B026-C05 | Almost all large front-runners are building new AI-era roles; five are described; two-thirds of the firms describe training people to direct AI. | Supported | www.bcg.com | 24 September 2026. Qualitative, 50-plus front-runners. Not prevalence. |
| B026-C06 | 6% of 600-plus large US public companies are AI leaders. Three-year industry-adjusted TSR vs the sample median: leaders +9.3 points, active tier +0.6, laggards -1.7. | Supported | www.bcg.com | 9 July 2026. Adoption tiers, not redesign. Annualized or cumulative not stated. Correlational. |
| B026-C07 | Firms that redesign work and talent outperform peers by "more than 11 percentage points" a year | Rejected by the Brief | www.bcg.com | 11.0 points is the leader-laggard spread over three years, not a gap over peers; no annual basis is stated. |
| B026-C08 | Organizational factors carry 67% of reported AI impact, individual mindset and behavior 32%. 45% say current goals feel safer than redesigning work with AI; 13% say they're rewarded for reinvention even if results aren't met. | Supported | www.microsoft.com | 5 May 2026. 20,000 AI users. Vendor survey. Self-reported. |
| B026-C09 | 69% of firms use AI, more at younger and more productive firms. 9 in 10 executives report no employment or productivity impact over the past three years. | Supported | www.nber.org | Working Paper 34836, February 2026, revised March 2026. Not peer reviewed. |
| B026-C10 | Redesign causing the premium, the five roles' prevalence, and redesign shares from pages not opened or undated | Withheld | - | Not established. |
| B026-C11 | Over the next 12 months, companies adding AI tools to unchanged processes keep trailing redesigners on reported business improvement, while front-runners keep adding roles to run AI work. | Autonoma synthesis | - | Autonoma join. Labeled forecast. |
Source Ledger
- [1] Goel, Bedard, Kropp, Sim, and Kaffe, "Five Ways That AI Front-Runners Change How Work Gets Done." BCG, 24 September 2026. Qualitative study of more than 50 front-runner firms, March to August 2026. Holds the five roles, the training share, and the rejected 11-point line. https://www.bcg.com/publications/2026/companies-use-ai-to-redesign-work
- [2] Pidun, Job, Wang, Molloy, Grebe, Franke, and Romanov, "AI Talk Is Cheap. Value Creation Is Rare." BCG Institute, 9 July 2026. Four adoption tiers across more than 600 US public companies above $5 billion. Holds 6%, +9.3, +0.6, and -1.7. https://www.bcg.com/publications/2026/how-ai-leaders-create-competitive-advantage
- [3] Beauchene, Duranton, Martin, Lyon, and Walters, "AI at Work: Strategy Matters More Than Tools." BCG, fourth edition, 3 June 2026. Survey of 11,749 people in more than a dozen markets. Holds the employer split and the training, clarity, role-shift, and guidance shares. https://web-assets.bcg.com/eb/92/4b39b729403fb6fcae4ff974b234/ai-at-work-slideshow-jun-2026-1.pdf Article: https://www.bcg.com/publications/2026/ai-at-work-why-strategy-matters-more-than-tools
- [4] Microsoft WorkLab, "Agents, human agency, and the opportunity for every organization." 2026 Work Trend Index, 5 May 2026. 20,000 AI-using knowledge workers, fieldwork ending April 2026. Holds 67%, 32%, 45%, and 13%. AI users only. https://www.microsoft.com/en-us/worklab/work-trend-index/agents-human-agency-and-the-opportunity-for-every-organization
- [5] Yotzov, Barrero, Bloom, Bunn, Davis, and others, "Firm Data on AI." NBER Working Paper 34836, issued February 2026, revised March 2026. Nearly 6,000 executives in the US, UK, Germany, and Australia. Holds 69% and 9 in 10. https://www.nber.org/papers/w34836
Unused here: Deloitte's undated State of AI 2026 shares, McKinsey's unopened State of AI 2026 page, a separate Microsoft study's manager effect, and other BCG and NBER shares.
Evidence Boundaries
- Self-classified employer types are not audited, so the Deploy, Reshape, and Invent gaps are associations.
- A three-year premium for an adoption tier is not a return on redesign, and the study does not say whether it is annualized.
- Front-runner interviews show what leaders built, not how common it is; "archetypes" is this Brief's word for BCG's roles.
- Microsoft's shares are reported impact among people who already use AI.
- Strong tool access is a low bar in BCG's terms, and three of the five sources are BCG's.
- No legal advice, and no finding about a named employer.
Dissent and Limiting Case
The strong objection is selection: AI use is higher at younger and more productive firms, so redesigners may be firms already ahead. Separately, 9 in 10 executives report no employment or productivity impact over the past three years. BCG's labels also look generous: 8,071 of its 10,990 respondents put their employer in Reshape or Invent.
The claim survives narrowed: the reported gap sits with redesign, and front-runners build roles and training around it. No source tests whether tools on unchanged processes deliver the same gains, or whether the gap survives controls for prior performance.
Falsification
This Brief is wrong, or has to be rewritten, if:
- the Reshape or Invent advantage disappears once prior firm performance is controlled; or
- companies that only deploy tools report the same impact and redesign participation as companies that redesign; or
- the new roles prove to be renamed existing jobs, with no change in the work.
It gets tighter if a study measures output, not self-report, before and after a redesign.
Forecast Label
The sentence "Over the next 12 months, companies that keep adding AI tools to unchanged processes will keep trailing redesigners on reported business improvement, while front-runners keep adding roles that design, govern, and supervise AI work," and its line on L&D programs built only on tool fluency, are Autonoma Intelligence synthesis. The timing is not a fact quoted from any source above.
Methodology
This audit checks Brief 026 against the public sources in the ledger, and it holds each figure to the page that states it. The join across the redesign gap, the new roles, and the training and incentive gaps, plus the 12-month timing, is Autonoma Intelligence synthesis. This draft was prepared with an AI assistant and reviewed and approved by the editor.