Topicsagentic-ai

Agentic AI in Manufacturing: ROI Claims Against Failure Rates

A sourced reference collection on autonomous AI agents in industrial operations — what vendors have actually shipped, what results they claim, and why the published success and failure rates cannot all be true at once.

Assertions
7
Sources consulted
6
Read in full
4/6
Cited as evidence
4

6 sources sit behind this page — including any that arrive with a concept this page shares with another collection. 4 were retrieved and read in full, and only those can back an assertion. 2 could not be retrieved. Every one of them is named in the register below, with the reason in view. How we source this.

Timeline newest first · evenly spaced, not to scale

NOWEARLIERMay 12, 2025Siemens Introduces Industrial AI Agents at Automate 2025
Background

Our own synthesis, written to orient you — not evidence. Every factual statement here is asserted and sourced further down this page.

This is the topic on this site where the published numbers most obviously cannot all be true at once, and the collection is built around that rather than around a resolution of it. Vendors publish case studies with large returns. Analysts forecast that a large share of agentic AI projects will be cancelled. Adoption surveys report near-universal experimentation. All three can be simultaneously accurate only if most deployments are small, most benefits are modest, and most ambitious projects fail — which is roughly what the neutral data suggests.

The neutral data comes from Stanford's AI Index, which compiled organisational AI use rising to 78% of surveyed respondents in 2024 from 55% a year earlier, and regular generative AI use in at least one business function more than doubling from 33% to 71%. Among business functions, the ones nearest to industrial operations lead on reported cost savings: supply chain and inventory management at 61%, service operations at 58%. The compilation's own qualification is the sentence to keep: those cost and revenue effects are reported “most commonly at low levels”.

Widespread adoption producing small measured effects is not a contradiction of either the vendor case studies or the cancellation forecasts. It is what you would expect if the easy wins are real but bounded, and the ambitious autonomous deployments — the ones that would justify the word agentic — are mostly still failing to reach production.

What is missing is named. The most-quoted figure in this subject, an analyst prediction that over 40% of agentic AI projects will be cancelled, reaches these pages only at one remove: the original press release returned HTTP 403 to the fetcher and, on a later attempt, a human-verification challenge that this project will not attempt to defeat. Its poll, sample size and the related vendor ‘agent washing’ estimate are absent entirely, and no assertion here rests on the figure alone.

Figures

Every number below is asserted and sourced elsewhere on this page.

Adoption climbed; the effects stayed small

Share of surveyed organisations using AI in at least one business function. The rise is real — the compilation that reports it also says the resulting cost and revenue effects are most commonly at low levels.

Surveyed organisations using AI, per cent

55%

2023

78%

2024

Stanford AI Index Report 2025, compiling McKinsey’s 2024 survey. Survey responses about organisational use, not audited deployments, and a consultancy’s sample compiled by an academic institute rather than an independent measurement.

Where the cost savings are reported

Share of respondents reporting cost decreases, by business function. The two functions closest to industrial operations lead — which is the strongest signal in this collection that the technology does something.

Total % of respondents, by function

  • Supply chain and inventory61%
  • Service operations58%
  • Human resources56%
  • Strategy and corporate finance56%

Stanford AI Index Report 2025, compiling McKinsey’s 2024 survey. These are shares of respondents reporting any cost decrease in that function, not the size of the decrease; the report states the effects are most commonly at low levels. The parts overlap and do not sum to a whole.

Concepts

The vocabulary this subject is built from, and what we can show about each.

Agentic AI

other

Industrial agentic AI executes workflows rather than answering questions — monitoring process data, deciding and acting — as distinct from conversational copilots; Siemens describes its agents as independently executing complete industrial workflows.

ReportedSupported by the sources below, not yet editor-reviewed.
1 source1 retrieved & read

At Hannover Messe 2026, twelve IoT Analytics analysts counted 29 industrial agentic AI solutions of which 72% were commercially available rather than pilots and 78% were model-agnostic — a measure of what vendors are selling, not what manufacturers are running.

ReportedSupported by the sources below, not yet editor-reviewed.
1 source1 retrieved & read

Agentic AI Project Outcomes

other

Gartner's widely-quoted prediction that over 40% of agentic AI projects will be cancelled by end-2027 reaches this collection only through trade coverage: Gartner's own release and a reproduction of it both returned HTTP 403, so its poll, sample and supporting estimates are not asserted here.

ReportedSupported by the sources below, not yet editor-reviewed.
1 source1 retrieved & read

Published agentic AI outcome figures are mutually incompatible: failure and cancellation rates of 40%, 50%, 76.4% and 80% are reported alongside a 171% average ROI and a 4-8 month payback, with no source explaining how both can describe the same population.

ReportedSupported by the sources below, not yet editor-reviewed.
1 source1 retrieved & read

Stanford's AI Index compiled organisational AI use rising to 78% of respondents in 2024 from 55% in 2023, with cost decreases most often reported in supply chain and inventory management (61%) and service operations (58%) — but qualifies that the reported effects are 'most commonly at low levels'.

ReportedSupported by the sources below, not yet editor-reviewed.
1 source1 retrieved & read

Industrial Copilot

other

Siemens' Industrial Copilot family spans NX CAD design, TIA Portal code generation, Insights Hub operations and Senseye maintenance, with a stated target of up to 50% customer productivity improvement and a reported 25% reduction in reactive maintenance time from Senseye pilots — all figures the vendor's own.

ReportedSupported by the sources below, not yet editor-reviewed.
2 sources2 retrieved & read

Timeline

What actually happened, in order, with sources.

Coverage
  • 1 United States

Where this topic’s events took place, as far as our sources establish it. Events with no single location — a standards publication, say — and events we have not yet attributed are both counted as unattributed rather than omitted.

  1. May 12, 2025

    Siemens Introduces Industrial AI Agents at Automate 2025

    technology generation milestoneUnited StatesNorth America

    On 12 May 2025 at Automate 2025 in Detroit, Siemens introduced industrial AI agents across NX CAD, TIA Portal, Insights Hub and Senseye within its Industrial Copilot ecosystem, stating a target of up to 50% customer productivity improvement and reporting a 25% reduction in reactive maintenance time from Senseye pilots.

    ReportedSupported by the sources below, not yet editor-reviewed.
    1 source1 retrieved & read

Source register

All 6 sources behind this page — what we read, what we tried to read and could not, and what we looked at and set aside, with the reason in view for each. A concept shared with another collection brings its own references with it, so some entries here were surfaced for a neighbouring topic rather than this one.

Cited as evidence
4
Tried, could not read
2
Surfaced, set aside
0
Cited sources 4 distinct links

Original publisher links. Files open on the publisher’s site; we do not host copies. A linked document is not an additional source or an independent verification.

Tried, could not read2

We attempted these and were refused or served nothing. Nothing on this page rests on them; they are published so the gaps are checkable rather than invisible.

Coverage & limits

What this page does and does not claim.

Tenth packet through the pipeline, rebuilt in September 2026 with a neutral baseline it previously lacked. (Original note, August 2026: Three sources were directly retrieved and two are recorded as references only: Gartner's press release and a trade reproduction of it both returned HTTP 403. That gap shapes the collection, because Gartner's over-40% cancellation prediction is the most-quoted figure in the subject and it reaches these pages only at one remove — its poll, sample size and the related 'agent washing' vendor estimate are absent entirely. The central assertion here is not a fact but an incoherence: reported failure and cancellation rates of 40%, 50%, 76.4% and 80% sit alongside a reported 171% average return on investment and a four-to-eight-month payback, in the same article, unreconciled. Both halves are recorded with their attributions and the contradiction stated. All quantified deployment results in this collection originate with the vendor that sells the product, and are labelled as targets or pilot figures rather than audited outcomes. Siemens and Rockwell Automation appear carrying placeholder descriptions inherited from an earlier development catalog; those are restated unchanged rather than overwritten, and need editorial correction. Not yet editor-reviewed; every assertion reads as reported.) Stanford's AI Index was added as the neutral counterweight to vendor case studies and analyst forecasts alike, and its own qualification that reported effects are most commonly at low levels is carried into the assertion rather than dropped. The Gartner gap remains: re-checked in a later session, that publisher serves a human-verification challenge which this project will not attempt to defeat, so the most-quoted figure in the subject is still held at one remove.

Source check, 2026-09-17. Numeric-presence checks passed for 7 assertions using available source text, which may be cached. This is not verification of their meaning. What this check does and does not prove →

  • Not editor-reviewed unless labelled. Assertions marked Reported are assembled from the sources shown and have not yet been checked by an editor. Only Primary source and Corroborated mean a human verified them.
  • Disagreements are preserved, not resolved. Where sources conflict, both accounts appear and the assertion is marked Disputed.
  • Retrieval status is disclosed per source. A source we could not open is never counted as evidence for an assertion.

This page is also available as structured data: /api/v1/topics/agentic-ai