Strategic PartnersIssue 117 records
Cisco
Cisco Live APJC (Melbourne) coverage tonight centers on constraints rather than outcomes. Independent reporting shows Cisco Cloud Control is still U.S.-only (Australia/EU/UK Feb. 2027), that customers allow autonomous network actions only for low-risk fixes, and that Splunk's agent-evaluation claims depend on customer-defined standards. An independent analyst note shows Cloud Control's Splunk integration is limited to Splunk Cloud on U.S. commercial AWS, and Cisco's own Omdia survey headline (51% agentic AI in production) conflicts with its per-process data and a 5% figure cited on stage. For research universities, the Cisco-backed NIIN sovereign AI cloud in Australia reached prototype with sensitive-data use cases but no outcomes. No new U.S. SLED Cisco AI deployment results were found.
- Evidence records
- 7
- Cross-source patterns
- 2
- Evidence classes
- 5 independent reporting1 independent research1 vendor claim
- Outcomes
- 3 emerging3 mixed1 cautionary
- Source freshness
- 5 new this fortnight1 older, newly relevant1 undated
- Research completed
- 2026-11-11
Choose a role to see its takeaway beside every record in the ledger.
Patterns across the evidence
Synthesis · Lighthouse Advisory interpretation
Agentic network-operations adoption claims diverge from the bounded autonomy customers actually grant
Cisco's commissioned Omdia survey headlines 51% running action-taking agentic AI in production, but its own process-level chart shows about 14-18% per process, and a Cisco executive cited only 5% deploying agents to production at the same event. Vaccaro says customers allow autonomous fixes for narrow, low-risk issues, and Moor Insights notes every Cloud Control change stops at a human approver. For SLED, adoption percentages are not a reliable basis for autonomy decisions; local baselines and graduated authority are.
Operating questionWhich specific network actions would you allow an agent to take without approval, and what local baseline will show whether that is working?
Supporting evidenceCisco with Omdia (commissioned survey)Network World / ZK Research (Zeus Kerravala), reporting on CiscoCisco (executive interview reported by iTWire)Moor Insights & Strategy (Mike Leone)
Cisco's agentic platform value is gated by regional availability, cloud stack and integration scope
Cloud Control is available for U.S. infrastructure only, with other regions in 2027; its Splunk path was limited to Splunk Cloud on U.S. commercial AWS (as of August); Tokenomics depends on AWS Bedrock/CloudWatch data and AWS Marketplace; Model Foundry is partner-first until June 2027. Agencies running Splunk Enterprise on-premises, government cloud tenants or non-AWS agent stacks may not receive the cross-domain capabilities in the keynote narrative.
Operating questionWhich of your data sources and agent platforms are connected to Cisco's agentic tooling today, rather than promised on a roadmap?
Supporting evidenceCisco (reported by iTWire)Moor Insights & Strategy (Mike Leone)Network World / ZK Research (Zeus Kerravala), reporting on CiscoSplunk (Cisco), reported by iTWire
Evidence ledger
Full record · every source keeps its link and limitations
iTWire: Cisco-backed NIIN Sovereign AI Cloud reaches live prototype with four university research use cases; production criteria, funding and access still undecided
iTWire reports that NIIN's Cisco-infrastructure Sovereign AI Cloud is now a live prototype (with partners Xenon and Sharon AI), with the University of Queensland announced as a second node. Four use cases test requirements: a UTS emergency-services AI/cyber testbed that currently trains on single-GPU desktops, Curtin mining hazard vision with identifiable images, RMIT/Austin Health de-identification of identifiable health records, and UQ enzyme discovery on commercially sensitive data. Production is the next phase; Cisco's Stefan Leitl said production criteria, funding and access arrangements will decide that move.
Why it matters, evidence and limitations
- Why it matters
- The model of a university-governed, shared GPU environment for sensitive research data parallels U.S. research-university and consortium compute (e.g., state R&E networks), and the use cases touch emergency services and health data. It is a prototype in Australia with no outcome data, so it supports discovery conversations about governance and sensitive-data enclaves, not claims of proven results.
- Evidence and measured results
- Trade-press article drawing on Cisco executive briefing and Cisco Live presentation. Describes requirements and use cases; reports no performance, cost, utilization or research-outcome metrics. Infrastructure specifics (GPU counts, architecture) are not disclosed.
- Limitations and uncertainty
- International prototype; no measured outcomes. Relies on vendor executive statements. U.S. transferability is limited by different research-data regulation (HIPAA, FERPA, state privacy laws), funding models and procurement rules.
Network World (ZK Research): Cisco Live Melbourne frames agentic AI as a network problem; Cloud Control still US-only, lock-in skepticism, conflicting readiness figures
Kerravala summarizes Cisco's Melbourne keynote: ThousandEyes telemetry added to Wi-Fi 6E/7 access points at no extra cost for eligible Networking Essentials/Advantage licenses; Cloud Control as the 'single front door', with Schultz saying no new Cisco product ships without Cloud Control integration; Cloud Control available only for US-based infrastructure, with Australia/EU/UK in February 2027 and Japan/India/Canada in May. Cisco executives cited IDC data that 9% of regional organizations feel networks are fully ready, and research that 85% experiment with agents but only 5% deploy to production. The analyst agrees with the thesis but says lock-in-concerned customers will be skeptical and the open-ecosystem story is unproven.
Why it matters, evidence and limitations
- Why it matters
- For U.S. state/local and campus networks the practical items are telemetry on new access points (a refresh-cycle decision), Cloud Control's growing role as a gate for new Cisco products, and the readiness gap between agent experimentation and production. The 5% production figure cited on stage contrasts sharply with Cisco/Omdia's 51% claim, which buyers should reconcile before accepting adoption narratives.
- Evidence and measured results
- Analyst commentary after keynote and executive interviews; figures (9%, 85%/5%, 22%→78% exploits on or before disclosure day, '98% lower cost' Eval-G1 claim) are attributed to Cisco executives or unnamed research, not independently verified. Customer stories (Asahi Beverages, XLSmart) are private-sector and APJC.
- Limitations and uncertainty
- Analyst provides services to vendors (disclosed bio describes advising vendors); no hands-on testing; statistics are second-hand from keynote; private-sector customer examples only.
iTWire: Cisco's Joe Vaccaro says customers permit autonomous agent actions only for low-risk fixes such as VLAN mismatches, port bounces and AP reboots; agents must show evidence
Cisco SVP Joe Vaccaro says customers already allow autonomous actions for issues such as VLAN mismatches, port bounces and access-point reboots, but more advanced autonomy carries a different risk profile, so agents must show the signals, evidence and workflow behind a fix. Cisco's AgenticOps sequence is sense, locate, plan, validate, deploy, using deterministic execution workflows from customer policies, digital-twin testing before changes and ThousandEyes checks after. He cites Cisco/Omdia research that 46% of network alerts are closed without investigation, and suggests engineers document expertise as Markdown skills while enforceable permissions and approvals remain separate.
Why it matters, evidence and limitations
- Why it matters
- Lean SLED network teams facing alert backlogs could start with the same low-risk autonomous fixes and require evidence trails before widening authority. The guidance on capturing staff knowledge as runbooks for agents is relevant to public-sector retirement and turnover risk.
- Evidence and measured results
- Vendor executive statements reported by trade press; no customer names, counts, error rates or measured outcomes. The 46% figure comes from vendor-commissioned Omdia research.
- Limitations and uncertainty
- Vendor executive opinion; 'customers' unnamed; digital-twin capability is described as what 'the system will use' (future tense). Omdia figure is vendor-commissioned.
iTWire: Cisco Cloud Control, GA in the US since Aug. 25, targets a sovereign Sydney instance in Feb. 2027 with five or six Australian beta customers; Model Foundry GA slated June 2027
iTWire reports Cloud Control reached U.S. general availability on August 25, 2026, with an Australian sovereign instance in Sydney targeted for February 2027 alongside the EU and UK, and Japan, India and Canada in May. Cisco is working with five or six Australian beta customers. ThousandEyes in Wi-Fi 6E/7 APs with eligible licenses targets general availability in December 2026; Experience Metrics became GA in Meraki Dashboard on Nov. 9. Model Foundry (controlled availability, partners first, optional fine-tuning on customer data) is slated for GA in June 2027. Schultz cited OpenRouter data that agentic API keys used about seven times the tokens of human keys in late September.
Why it matters, evidence and limitations
- Why it matters
- U.S. SLED buyers can use Cloud Control now on U.S. infrastructure, but timelines show several agentic features (ThousandEyes default, Model Foundry) are not yet generally available; fine-tuning on customer data via Model Foundry raises data-governance questions for public records and network configurations.
- Evidence and measured results
- Trade-press reporting of keynote statements and an executive interview; dates are vendor targets. No customer outcomes.
- Limitations and uncertainty
- Timelines are targets; the article mixes 'Cisco Live Melbourne 2027' wording with 2026 dates (editorial inconsistency). Australian focus; U.S. state data-residency needs are not addressed. OpenRouter token data is a third-party service sample, not network telemetry.
iTWire: Splunk Tokenomics and Eval-G1 aim to show whether AI agents earn their keep; usefulness depends on customer-defined standards and evaluator performance
iTWire describes Splunk's Tokenomics (cost-to-outcome view) and Eval-G1 small evaluator models with Evaluation Tuner. It reports Splunk's claims of 2.5x faster responses than LLM judges and accuracy of up to 95%, and Cisco's claim of up to 98% lower evaluation cost, while noting usefulness depends on the standards supplied and evaluator performance on actual work, and that customers must define what makes work worthwhile. It cites Splunk's State of Observability 2026 (Oxford Economics, 2,050 IT operations/engineering professionals in 10 countries, Feb–Mar 2026): 55% could not control AI costs; 60% said billing spikes caused more strain than baseline costs. Agent Launchpad access via Codex and Claude Code is scheduled for November.
Why it matters, evidence and limitations
- Why it matters
- SLED agencies piloting agents face budget scrutiny and accountability for AI outputs; cost-to-quality measurement is relevant, but value depends on agencies defining their own quality standards. Tokenomics' AWS Marketplace and Bedrock/CloudWatch dependence limits immediate fit for non-AWS agencies.
- Evidence and measured results
- Trade press relaying vendor claims with explicit independent caveats; survey is vendor-commissioned (Oxford Economics) and not public-sector specific; benchmark methodology for the 95% accuracy figure is not disclosed.
- Limitations and uncertainty
- Accuracy and cost claims are vendor-stated; 'up to' qualifiers; survey not public-sector specific; preview/GA timing per vendor.
Moor Insights & Strategy: Cisco Cloud Control's cross-domain agent pitch depends on sources still being connected; Splunk integration limited to Splunk Cloud on US commercial AWS stacks
The analyst finds Cloud Control's agents stage fixes for human approval and that AI Canvas gives teams shared evidence, but cautions that the cross-domain argument depends on sources still being wired in: Splunk connects only via a recent Splunk Cloud Platform release on U.S. commercial stacks hosted in AWS, and telemetry in other Splunk products stays outside AI Canvas. Controlled availability covered only U.S. infrastructure at the time. The note recommends asking which data sources are connected today rather than promised, what the platform can conclude about estates the vendor does not own, and what operators see before approving production changes.
Why it matters, evidence and limitations
- Why it matters
- Why now: Cisco's Nov. 10 Melbourne keynote re-centered Cloud Control as the 'single front door' and gate for new products and announced regional rollout dates, making these integration limits current. Interpretation: many SLED agencies run Splunk Enterprise on-premises or in government cloud environments, which this note indicates would not feed AI Canvas, so cross-domain agent claims may not apply to their estates.
- Evidence and measured results
- Analyst research note based on vendor briefings and product documentation; cites EY Technology Pulse Poll (78% of technology leaders say AI adoption outpaces risk management). No hands-on test results.
- Limitations and uncertainty
- Published August 2026; integration scope may have expanded since (Cloud Control reached U.S. GA Aug. 25 per later reporting). Analyst firm may have vendor relationships. Not public-sector specific.
Cisco/Omdia 'The Impact of Agentic AI on Network Operations' (2026): vendor-commissioned survey claims 51% run agentic AI in production NetOps, yet per-process use is 14–18%
The report states Omdia surveyed over 1,000 IT and network operations leaders at organizations with 500+ employees across North America, Western Europe and Asia-Pacific, supplemented by Cisco Meraki telemetry analysis. It claims 51% are running agentic AI that takes action in production and 84% expect an AI-led operating model within 12 months, while its own process-level chart shows only about 14–18% using agentic AI today for any single NetOps process (e.g., 18% remediation actions). It reports 46% of network alerts are closed without investigation that 67% say the generative AI boom significantly increased network complexity, and that 70% say AI adoption increased NetOps workloads.
Why it matters, evidence and limitations
- Why it matters
- Why now: Cisco re-cited this survey at Cisco Live Melbourne (Nov. 9–11 blog and Vaccaro interview) to justify Agentic Actions, while a Cisco executive on stage cited separate research that only 5% have deployed agents to production. Interpretation: SLED buyers should treat the 51% headline as a broad, vendor-commissioned definition and benchmark locally; the sample excludes organizations under 500 employees and reports no public-sector breakdown.
- Evidence and measured results
- Survey of 1,000+ respondents; question wording, sector mix and public-sector share not disclosed in the report; commissioned by the vendor whose product it supports.
- Limitations and uncertainty
- Vendor-commissioned; no public-sector breakdown; definitional gap between headline 51% and process-level 14–18%; undated PDF, so publicationDate is null.
How to read this edition
Source findings, measured results and limitations come from the cited publications. Patterns, operating questions, role takeaways and implementation considerations are Lighthouse Advisory interpretation, stated as questions to validate locally rather than guaranteed outcomes. Vendor and operator claims are labeled as claims. Full research method.
- Independent reporting
- Independent reporting with attributable sources but without a formal evaluation design.
- Independent research
- Research conducted outside the implementing organization.
- Vendor claim
- A supplier-provided assertion that has not been upgraded to independent evidence.