Appetite for SAIA 2.0 + Agent Mode among these 54 mostly-admin respondents is real but conditional: the gate is governance and data control, not desire. Data-privacy/residency (41 of 51 flagged it) and verifiable accuracy have to be cleared before tepid 90-day intent (20 of 52 likely or very likely) converts, so the next move is a trust-and-control package, not more feature breadth.

Confidence: Moderate-Thin Generated 2026-06-20 01:23 UTC · n=54 real · 54 real responses survive the is_test filter (0 flagged); 52 are substantive completes and 2 are near-empty drop-offs. The sample is a self-selected, AI-curious convenience sample skewed toward Splunk Platform Administrators (16) and 'Other' (10); several role slices are single digits (DevOps 2, Data Engineer 1, Power User 1), so whole-sample patterns are directional only and no per-role claim is defensible. Unweighted, single survey instance.

Key insights

1Data residency and control is the hard gate: 41 of 51 flagged data-privacy and 25 of 51 flagged data leaving the perimeter, and the open-text repeats on-prem / airgap / data-sovereignty as a binary no-go for regulated and public-sector respondents.

2Intent is tepid and conditional, not enthusiastic: only 20 of 52 are likely or very likely to activate within 90 days (19 neutral, 13 unlikely) and just 9 of 47 report having activated, yet 47 of 52 are already aware, so the gap is not ignorance.

3Trust means accuracy you can verify, benchmarked against Claude/ChatGPT: 32 of 52 cite a trust/mental-model gap, and 28 of 54 already use a frontier model while only 14 use the Splunk AI Assistant, so SAIA has to beat the tool already open in the next tab.

Recommendation: Lead the next SAIA 2.0 roadmap and go-to-market with a governance-and-trust package: a documented data-flow/residency story (including an on-prem or bring-your-own-model path), admin controls (RBAC scoping, full audit trail, human-in-the-loop approval, cost caps), and environment-aware SPL with cited source events, before further feature breadth. Treat this as directional given the self-selected, admin-heavy sample, and validate residency and cost thresholds with the actual approvers (CISO / compliance / finance) before committing.

Pass 2 — hypothesis check

#Working hypothesisVerdictSupportingContradictoryConf.H1There is meaningful stated intent to activate Agent Mode within 90 days.Partially supported20 of 52 are likely or very likely to activate in 90 days; 24 of 52 expect it to change their work 'a lot' or more; awareness is high at 47 of 52; one manager wrote 'Once approved and activated there would be immediate and enthusiastic adoption.'The plurality is neutral (19 of 52) and 13 of 52 are unlikely or very unlikely; only 9 of 47 report having activated. Intent is conditional ('it all depends after validating'), not a committed pipeline.Moderate-ThinH2Organization-level blockers (risk, governance, data, cost) are distinct from and dominate user-level blockers (awareness, trust, skill).SupportedOrg blockers cluster on data privacy (41 of 51), compliance (30), cost visibility (29), and perimeter (25); user blockers cluster on the trust/mental-model gap (32 of 52), onboarding/discoverability (28), and accuracy (22). The open-text shows the same split: orgs talk RBAC/residency/cost, users talk 'not knowing it's there' and training.The boundary blurs at the edges: cost surfaces as a user-level blocker too ('I need complete understanding of the cost implications before I can ask the team to use it'), and accuracy/trust spans both levels, so the two registers overlap rather than being cleanly separate.Moderate

The ATX UXR Blog

/Users/maralelliott/Downloads/0-Enablement/_HANDOFF-1cd-redesign/

Don't miss out! Join our mailing list.