Rate your overall agentic AI expertise

The mean rating was 4.9 and the median 5.0 (n = 800 respondents). Ratings of 3 or below, the distressed range, made up 27.75% of the answers. The rating was associated with all four study measures. Higher ratings came with less cancellation (r = −0.61, 95% CI −0.70 to −0.50, p < 0.001, n = 150 engagements), higher customer satisfaction among active engagements (r = 0.74, 95% CI 0.63 to 0.82, p < 0.001, n = 94 engagements), a higher engagement indicator mean (r = 0.78, 95% CI 0.71 to 0.84, p < 0.001, n = 150 engagements) and lower intent to leave (r = −0.78, cluster bootstrap 95% CI −0.82 to −0.73, p < 0.001 with standard errors clustered by engagement, n = 753 respondents). After Holm correction, ratings differed by engagement status and lifecycle stage (Welch F 34.84, Holm p < 0.001, omega squared 0.363) and respondent role (Welch F 9.52, Holm p = 0.001, omega squared 0.018).

The 37 written responses came from C030, on what the respondent loves most about the job (30), and C135, the comment field on the employer's capability (7). Twenty-four were positive and four negative. Every response to C030 named agentic AI skill learned on the engagement, most often the quality assurance of agent output. The responses to C135 described training on the agents of days or videos, and respondents who knew more about AI than those above them. No written outcome test reached significance.

Ratings of 3 or below, the distressed range, accounted for 222 of the 800 answers (27.75%). The engagement mean fell in the distressed range in 26 of 150 engagements.

All 37 responses were secondary. Each answered a question outside the indicator's survey section and addressed the indicator in its content. They came from 2 questions. That secondary share ranked 1st from the highest among the 11 Part 4 indicators with 30 or more written responses, tied with seven other indicators, against a median of 100.00%.

Each of the thirty responses to C030 listed agentic AI skill learned on the engagement among the things the respondent loves most about the job. Fourteen named the quality assurance or evaluation of agent output, often for agents that do not give the same answer twice, and several said few peers had that skill. Others named observability at an agent's boundary, identity and access provisioning for agents, production operations and building agentic playbooks. Eight set the learning beside an engagement that stalled, sank, fell apart or was canceled or terminated.

The seven responses to C135 described the employer's expertise. Three described training on the agents of two days of slides, a two-week course or two videos, and two said no one above the respondent had run such agents before, one adding that the work had been sold as if they had. Three respondents early in their careers said they knew more about AI than their managers, their trainer or the executive team, and one said a proposal to improve the AI practice went unread for lack of experience. Another contrasted the engagement, where each pod invented its own process for exceptions, with an earlier account that had one signed-off checklist.

Of the 37 responses, 4 (10.81%, 95% CI 2.50% to 22.23%) were negative and 1 (2.70%) carried the strongest negative score of −10. Positive responses numbered 24 (64.86%), and 9 were neutral. The mean written score was 2.6 (95% CI 1.2 to 3.7, n = 37 responses), and the median was 5.0.

Of the 37 responses, 32 came from respondents who also rated the indicator. Among them, the written score did not track the rating (Spearman rho −0.21, 95% CI −0.55 to 0.18, p = 0.243, n = 32 responses). Of the 2 negative responses with a rating, 1 came from respondents whose rating was in the distressed range.

Respondents who wrote about the indicator and gave a rating numbered 32, with a mean rating of 5.3. The 768 respondents who rated the indicator without writing about it had a mean of 4.9. The difference, 0.4 (95% CI −0.3 to 1.0), included zero. Ratings in the distressed range made up 15.63% of the ratings from respondents who wrote (5 of 32) and 28.26% of the ratings from respondents who did not (217 of 768).

Every figure in the segment tables is at respondent level. Segments with fewer than 30 respondents carry no interval. Segments with fewer than 10 report counts only.

Segment families recorded at engagement level were tested on engagement means. Respondent role and vendor type were tested on respondents. Where an expected count fell below 5, the test of the distressed share used Fisher's exact test for a two-by-two table, marked exact, or a Monte Carlo test, marked simulated. After Holm correction, ratings differed by engagement status and lifecycle stage (Welch F 34.84, Holm p < 0.001, omega squared 0.363). Active engagements in Production rated highest (mean 5.9, n = 98 respondents) and canceled engagements lowest (mean 3.5, n = 308 respondents). Ratings also differed by respondent role (Welch F 9.52, Holm p = 0.001, omega squared 0.018). Portfolio directors rated highest (mean 6.1, n = 31 respondents) and individual contributors lowest (mean 4.8, n = 617 respondents). Outsourcing model, institution type, process family, class of work and vendor type showed no difference.

Written sentiment did not differ by engagement status and lifecycle stage across the 32 engagements with a written response (Kruskal-Wallis H 2.74, df 3, p = 0.433). Of the groups with 10 or more responses, responses from active engagements in Transition/Cut-Over were the most negative (mean 2.7, 15 responses) and responses from canceled engagements the least negative (mean 2.9, 12 responses). Responses from active engagements in Build-Out (8) and active engagements in Production (2) are reported as counts only.

No test across respondent roles was run, because fewer than two roles had 30 or more written responses. Individual contributors wrote 30 and engagement managers wrote 7 responses.

Cancellation rate is the share of engagements canceled by the study date, measured at engagement level (150 engagements, 56 canceled). Canceled engagements were rated after cancellation. The association is therefore a descriptive contrast and not a forecast of cancellation.

Across the study set, higher ratings came with less cancellation (r = −0.61, 95% CI −0.70 to −0.50, Holm p < 0.001, n = 150 engagements). The association held after Holm correction in fourteen of the 20 segment levels tested.

Customer satisfaction is the percentage recorded against each engagement. The primary estimate uses the 94 active engagements. A descriptive contrast uses all 150 engagements.

Across the active engagements, higher ratings came with higher customer satisfaction (r = 0.74, 95% CI 0.63 to 0.82, Holm p < 0.001, n = 94 engagements). The association held after Holm correction in fifteen of the 16 segment levels tested.

The engagement indicator mean is the mean of an engagement's ratings across the rated indicators, measured at engagement level (150 engagements). The indicator does not enter the mean.

Higher ratings came with a higher engagement indicator mean (r = 0.78, 95% CI 0.71 to 0.84, Holm p < 0.001, n = 150 engagements). The association held after Holm correction in twenty-four of the 24 segment levels tested.

Intent to leave is the answer to C019, “How likely are you to leave your current job in the next 12 months?”, rated 1 to 10. It is measured at respondent level on up to 798 respondents, with the engagement-level form reported alongside.

Respondents who gave higher ratings reported lower intent to leave (r = −0.78, 95% CI −0.80 to −0.75, Holm p < 0.001, n = 753 respondents). The association held after Holm correction in thirty-seven of the 37 segment levels tested.

Two comparisons set the engagements with at least one negative response against all other engagements. Of the 146 other engagements, 118 had no written response addressing the indicator. These two comparisons therefore do not isolate written sentiment. Of the 4 engagements with a negative response, 2 were canceled, against 54 of the 146 other engagements (36.99%). The difference did not reach significance (Fisher exact p = 0.629). The comparison of customer satisfaction was not computed, because one of the two groups held too few active engagements.

Two correlations test written sentiment directly. Across the 32 engagements with a written response, the engagement mean written score was not associated with cancellation (point-biserial r −0.02, p = 0.916). Across 33 respondents, the respondent mean written score was not associated with intent to leave (Spearman rho 0.00, p = 1.000). Of the three comparisons, none reached significance at 0.05.

The four associations were compared in pairs on their engagement-level forms, each oriented so that a higher value is the better outcome, across the 150 engagements with all four measures. After Holm correction, five of the six pairs differed. The association with the engagement indicator mean was the strongest and the association with cancellation the weakest. The associations with customer satisfaction and the engagement indicator mean did not differ (Holm p = 0.200).

Answers clustered within engagements (ICC 0.63). Clustering reduced the 800 respondent answers to the equivalent of about 213 independent answers. Respondent-level intervals therefore used the cluster bootstrap. Respondent-level tests of intent to leave used standard errors that allow for clustering by engagement.

No association was equivalent to zero. Each of the four differed from zero. The detection floor shown is the study floor at 150 engagements. Intent to leave was tested on its effective sample size.

Each change was computed before rounding and may differ by 0.01 from the difference between the two printed values.

All 22 checks held. A check holds when its test falls on the same side of 0.05 as the primary test. Where the primary test falls below 0.05, the check must also keep the sign of the primary correlation. The thin-base check removes the engagements in any level of outsourcing model or process family with fewer than 30 engagements. Institution type is not used for that check, because every one of its levels has fewer than 30 engagements.

Each of the four measures was computed from data the indicator does not contain. Each result is therefore a finding and not an arithmetic consequence.

Each deviation was computed before rounding and may differ by 0.01 from the difference between the two printed values.

The comparison covered 150 engagements. The indicator departed from the study's typical pattern on intent to leave. Its association was stronger than the norm on intent to leave.

Among the 11 Part 4 indicators with 30 or more written responses, the indicator's share of negative responses, 10.81%, ranked 9th from the highest, against a median of 33.33%. Its mean written score, 2.6, ranked 7th from the most negative, against a median of 0.8.

The indicator most often present in the same responses was C080, “My manager is effective.”, in 8 of the 37 responses (21.62%). C036, “What stage is the engagement in?” (5 responses), C017, “Where did you gain your AI knowledge?” (4 responses), C123, “Rate the adequacy of your employer's investment in training its workforce.” (4 responses) and C157, “Rate the factual reliability of the AI tools' outputs on this engagement before any human review.” (4 responses) followed.

The most frequent theme was “Skills learned”, in 30 responses (81.08%). The second was “Team or colleagues”, in 18 responses.

Over a quarter of all ratings respondents gave their own agentic AI expertise fell in the distressed range (222 of 800 answers, 27.75%). The rating was associated with all four study measures, and every association held under each of the 22 robustness checks. Canceled engagements rated far below the three active stages (mean 3.5, n = 308 respondents).

The written evidence describes agentic expertise built on the engagement rather than brought to it, with the quality assurance of agent output as the skill most often gained. Respondents valued that skill even where the engagement failed. The employer's training on the agents was brief, and in several accounts the respondents' knowledge exceeded that of those above them. No written outcome test reached significance.

The evidence is the reported experience of the vendor's delivery teams, collected in a single cross-section. The associations reported here are not causal.

Tests by respondent role and by vendor type treated respondents as independent. The section on clustering reports how far clustering within engagements reduced the information in respondent-level figures.

Customer management executives gave 2 answers, a base of anecdote strength.

Shares of written sentiment are shares of responses and not of the study set of respondents. The written score takes five fixed values, so its mean and median are coarse. Responses clustered within engagements (37 responses on 32 engagements). Intervals for written sentiment therefore used a bootstrap by engagement.

The rating is a self-assessment of the respondent's own expertise and not a measure of it.