https://sled.lighthouseadvisory.consulting/feeds/cisco.xml2026-11-11T05:15:35.844ZLighthouse Advisoryhttps://sled.lighthouseadvisory.consulting/streams/cisco/editions/2026-11-102026-11-11T05:15:35.844Z2026-11-10T00:00:00.000Z

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.

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. 7 sources · 2 cross-source patterns Agentic network-operations adoption claims diverge from the bounded autonomy customers actually grant Which specific network actions would you allow an agent to take without approval, and what local baseline will show whether that is working? Cisco's agentic platform value is gated by regional availability, cloud stack and integration scope Which of your data sources and agent platforms are connected to Cisco's agentic tooling today, rather than promised on a roadmap?https://sled.lighthouseadvisory.consulting/streams/cisco/editions/2026-11-092026-11-10T07:05:30.432Z2026-11-09T00:00:00.000Z

Cisco used Cisco Live APJC (Nov. 9 U.S. time) to announce Anthropic Claude Mythos inside AI Defense adaptive red teaming (registered targets, named operators, audit trails; coming soon for select enterprises), Splunk Tokenomics ROI for AWS-hosted agents (preview December, GA early 2027), and Cisco IQ exposure data flowing into Splunk. Independent coverage (The Register, CRN) adds that Mythos-backed testing is planned under existing licenses and via partners, with roughly 400 to 500 AI Defense customers, but there is no independent testing or public-sector reference. Splunk's own documentation shows Tokenomics costs are vendor-reported and unaudited, and federal reporting from June shows government policy for Mythos-class tools lagging access. The one SLED deployment item is a Welsh county council's Webex contact-center consolidation (175,000 calls/month, 76% answer rate, vendor-reported) that is only beginning to explore AI agents. No new U.S. SLED Cisco AI deployment results were found.

Cisco used Cisco Live APJC (Nov. 9 U.S. time) to announce Anthropic Claude Mythos inside AI Defense adaptive red teaming (registered targets, named operators, audit trails; coming soon for select enterprises), Splunk Tokenomics ROI for AWS-hosted agents (preview December, GA early 2027), and Cisco IQ exposure data flowing into Splunk. Independent coverage (The Register, CRN) adds that Mythos-backed testing is planned under existing licenses and via partners, with roughly 400 to 500 AI Defense customers, but there is no independent testing or public-sector reference. Splunk's own documentation shows Tokenomics costs are vendor-reported and unaudited, and federal reporting from June shows government policy for Mythos-class tools lagging access. The one SLED deployment item is a Welsh county council's Webex contact-center consolidation (175,000 calls/month, 76% answer rate, vendor-reported) that is only beginning to explore AI agents. No new U.S. SLED Cisco AI deployment results were found. 6 sources · 2 cross-source patterns Frontier cyber-model testing reaches public-sector buyers through a vendor wrapper before agencies have policy for it Before any Mythos-backed or partner-run AI red-team test touches your systems, who approves it, which subprocessors handle the test data, and who will remediate what it finds? Agent ROI dashboards are being marketed while the underlying cost data remains vendor-reported and unaudited Which AI spend figures will you reconcile to invoices, and which outcome KPI will you define, before any agent ROI dashboard is cited in a budget request?https://sled.lighthouseadvisory.consulting/streams/cisco/editions/2026-11-062026-11-07T05:48:24.306Z2026-11-06T00:00:00.000Z

Nov. 6 Cisco IT published a cisco-on-cisco AI factory story claiming GPU utilization rose from ~15% to ~50% (peaks ~80%) while hosting MyAgent for 90,000 employees and Cisco IQ production, with AI Defense, planned Hypershield, Live Protect and Splunk observability in the stack—self-reported and unfinished on Tokenomics. The same research window’s Nov. 5 Splunk GSO Customer Zero post reports SOC triage handle time falling 15.4→5.6 minutes (64%) and detection authoring from ~2 hours to under 10 minutes under delegated/guided/bounded autonomy, with cloud ES tier gates and an alpha workflow builder called out. A Nov. 5 Webex customer story details Saudi HRSD’s in-house migration to Webex Suite/Calling (11,000+ users; 662,900 meetings; Virtual Branch citizen services), useful as international government collaboration evidence with explicit U.S. transferability limits. Anton Chuvakin’s Nov. 5 firewall essay reframes enforcement around agent task and delegated authority as AI traffic moves inward. Independent scrutiny comes from AI Security Platform’s Nov. 3 documentation review (7.6/10) newly paired against the AI Defense push: strong framework mappings and MCP/BOM coverage, but SSE dependency for workforce controls, private-offer pricing and no lab benchmark. All quantitative outcome claims tonight are vendor-reported except the docs review’s qualitative gap analysis.

Nov. 6 Cisco IT published a cisco-on-cisco AI factory story claiming GPU utilization rose from ~15% to ~50% (peaks ~80%) while hosting MyAgent for 90,000 employees and Cisco IQ production, with AI Defense, planned Hypershield, Live Protect and Splunk observability in the stack—self-reported and unfinished on Tokenomics. The same research window’s Nov. 5 Splunk GSO Customer Zero post reports SOC triage handle time falling 15.4→5.6 minutes (64%) and detection authoring from ~2 hours to under 10 minutes under delegated/guided/bounded autonomy, with cloud ES tier gates and an alpha workflow builder called out. A Nov. 5 Webex customer story details Saudi HRSD’s in-house migration to Webex Suite/Calling (11,000+ users; 662,900 meetings; Virtual Branch citizen services), useful as international government collaboration evidence with explicit U.S. transferability limits. Anton Chuvakin’s Nov. 5 firewall essay reframes enforcement around agent task and delegated authority as AI traffic moves inward. Independent scrutiny comes from AI Security Platform’s Nov. 3 documentation review (7.6/10) newly paired against the AI Defense push: strong framework mappings and MCP/BOM coverage, but SSE dependency for workforce controls, private-offer pricing and no lab benchmark. All quantitative outcome claims tonight are vendor-reported except the docs review’s qualitative gap analysis. 5 sources · 2 cross-source patterns Cisco and Splunk Customer Zero posts publish measured agent/infrastructure gains—still self-reported, still useful as pilot templates Which single SLED workload will you baseline this month so vendor Customer Zero percentages are replaced by your own evidence? AI Defense is being embedded in Cisco’s AI factory and trust story, while independent docs review and firewall essay demand honest scope—including SSE and task-aware network enforcement Is tonight’s AI Defense discussion about app/agent runtime, workforce shadow AI (SSE), network task enforcement—or all three conflated?https://sled.lighthouseadvisory.consulting/streams/cisco/editions/2026-11-052026-11-06T11:55:55.469Z2026-11-05T00:00:00.000Z

Cisco’s Partner Summit recap (Nov. 5) and a Splunk accountability blog (Nov. 4) push the same trusted-AI story: close the agent trust gap with security (AI Defense / Data Fabric), Splunk Agent Observability guardrails, and tokenomics so spend ties to outcomes. Independent iTWire coverage the same day restates ANZ customer hesitation and details Tokenomics, Galileo evaluation and AI POD for Splunk, with a vendor-told bank accuracy anecdote that is not audited. Against that pitch, Checkable AI’s Oct. 19 public-docs stack check—newly relevant tonight—finds AI Defense supports MCP runtime profiles and SOC export but not a complete tamper-evident per-agent action audit or public FedRAMP/IRAP/ISO 42001 attestation. No new U.S. SLED deployment results surfaced.

Cisco’s Partner Summit recap (Nov. 5) and a Splunk accountability blog (Nov. 4) push the same trusted-AI story: close the agent trust gap with security (AI Defense / Data Fabric), Splunk Agent Observability guardrails, and tokenomics so spend ties to outcomes. Independent iTWire coverage the same day restates ANZ customer hesitation and details Tokenomics, Galileo evaluation and AI POD for Splunk, with a vendor-told bank accuracy anecdote that is not audited. Against that pitch, Checkable AI’s Oct. 19 public-docs stack check—newly relevant tonight—finds AI Defense supports MCP runtime profiles and SOC export but not a complete tamper-evident per-agent action audit or public FedRAMP/IRAP/ISO 42001 attestation. No new U.S. SLED deployment results surfaced. 4 sources · 2 cross-source patterns Cisco and Splunk frame agent scale-up as a trust-and-cost problem: observe behavior, attribute token spend, then earn autonomy with evidence For which named agent workflow do we already have a cost-per-successful-outcome baseline and an interveneable guardrail path before we expand its authority? Trusted-AI marketing leans on AI Defense, while an independent docs check still finds incomplete agent audit and public attestation gaps Which agent-governance controls must AI Defense itself satisfy, and which will we meet with identity, PAM, SIEM retention or change-control systems instead?https://sled.lighthouseadvisory.consulting/streams/cisco/editions/2026-11-042026-11-05T05:43:52.925Z2026-11-04T00:00:00.000Z

Cisco focused on how AI agents get access. THE Journal reported Privileged Infrastructure Access, which gives admins, agents and scripts short-lived, recorded, task-scoped access to network devices using ISE and TACACS+ policy; the article notes no performance data. A Cisco blog says Cisco IQ now offers REST APIs and an MCP server where agents inherit caller permissions, with vendor scripted tests cutting tool calls from 256 to 36. Cisco also made a Cloud Intelligence Add-on for on-prem UCM generally available. No new SLED deployments surfaced.

Cisco focused on how AI agents get access. THE Journal reported Privileged Infrastructure Access, which gives admins, agents and scripts short-lived, recorded, task-scoped access to network devices using ISE and TACACS+ policy; the article notes no performance data. A Cisco blog says Cisco IQ now offers REST APIs and an MCP server where agents inherit caller permissions, with vendor scripted tests cutting tool calls from 256 to 36. Cisco also made a Cloud Intelligence Add-on for on-prem UCM generally available. No new SLED deployments surfaced. 3 sources · 1 cross-source patterns Cisco frames AI agent access as an identity problem: agents inherit scoped, auditable permissions rather than separate side doors Under which named identity, and with what task-scoped permissions, would we let an AI agent read our IT data or touch our network devices?https://sled.lighthouseadvisory.consulting/streams/cisco/editions/2026-11-032026-11-04T15:01:47.424Z2026-11-03T00:00:00.000Z

A light Cisco night. OpenAI published a Cisco customer story on Nov. 3 in which Jeetu Patel calls Daybreak and GPT-6 Astra step-function security improvements. It gives no metrics, and the joint Daybreak Defense Network is still described as exploratory. In a Nov. 3 TechTarget interview, Cisco said its network agents diagnose with AI but make changes only through predefined, deterministic workflows. A June TechTarget analysis supplies the counterweight: trust was the top barrier for 59.7% of surveyed respondents. No new SLED deployment results surfaced.

A light Cisco night. OpenAI published a Cisco customer story on Nov. 3 in which Jeetu Patel calls Daybreak and GPT-6 Astra step-function security improvements. It gives no metrics, and the joint Daybreak Defense Network is still described as exploratory. In a Nov. 3 TechTarget interview, Cisco said its network agents diagnose with AI but make changes only through predefined, deterministic workflows. A June TechTarget analysis supplies the counterweight: trust was the top barrier for 59.7% of surveyed respondents. No new SLED deployment results surfaced. 3 sources · 1 cross-source patterns Cisco and independent analysis converge: AI may diagnose network issues, but execution should stay in deterministic, reversible workflows Which network change classes are we willing to let an agent execute from a runbook, and what local evidence would justify expanding that list?https://sled.lighthouseadvisory.consulting/streams/cisco/editions/2026-11-022026-11-03T11:43:33.567Z2026-11-02T00:00:00.000Z

Tonight's Cisco edition has two Cisco announcements and two independent checks. Cisco says SnortML, the machine-learning engine in Firewall 10.1, blocked a new PHP2Shell exploit attempt within a day of disclosure, and Splunk has more than 10 post-exploit detections, though it gives no numbers. Wiz telemetry shows captured exploit attempts failing but about half of affected organizations still exposed after a week, so patching stays first. Cisco also introduced AI Transit to extend SD-WAN policy to specialized AI clouds, without availability or performance data. Futurum says Project Quarry agent detection is only as good as its rules and agent logs. No new SLED deployment results surfaced.

Tonight's Cisco edition has two Cisco announcements and two independent checks. Cisco says SnortML, the machine-learning engine in Firewall 10.1, blocked a new PHP2Shell exploit attempt within a day of disclosure, and Splunk has more than 10 post-exploit detections, though it gives no numbers. Wiz telemetry shows captured exploit attempts failing but about half of affected organizations still exposed after a week, so patching stays first. Cisco also introduced AI Transit to extend SD-WAN policy to specialized AI clouds, without availability or performance data. Futurum says Project Quarry agent detection is only as good as its rules and agent logs. No new SLED deployment results surfaced. 4 sources · 1 cross-source patterns Cisco's PHP2Shell ML-blocking claim is a compensating control; independent telemetry says patching and measured exposure tracking remain the main defense How many internet-facing PHP hosts remain unpatched, and is SnortML actually enabled and alerting into Splunk for them?https://sled.lighthouseadvisory.consulting/streams/cisco/editions/2026-11-012026-11-03T03:21:57.317Z2026-11-01T00:00:00.000Z

Tonight’s Cisco edition covers one new piece of Cisco AI security research. Cisco Talos found an infostealer, VOLTA, that sends stolen passwords, session tokens and personal data to commercial AI models to rank which victims are worth pursuing, falling back to other models if one refuses. Independent reporting from WIRED adds that Talos did not track whether the malware has been used in the wild. For agencies, campuses and school districts, the practical response is to block headless-browser abuse, watch AI API traffic and revoke stolen tokens faster. No new U.S. SLED Cisco AI deployment results surfaced.

Tonight’s Cisco edition covers one new piece of Cisco AI security research. Cisco Talos found an infostealer, VOLTA, that sends stolen passwords, session tokens and personal data to commercial AI models to rank which victims are worth pursuing, falling back to other models if one refuses. Independent reporting from WIRED adds that Talos did not track whether the malware has been used in the wild. For agencies, campuses and school districts, the practical response is to block headless-browser abuse, watch AI API traffic and revoke stolen tokens faster. No new U.S. SLED Cisco AI deployment results surfaced. 2 sources · 1 cross-source patterns Talos shows attackers using commercial AI to triage stolen data, but prevalence is unknown, so SLED defenders should shrink the window for using stolen tokens rather than react to a trend How long does a stolen staff or student session token remain usable in our environment today, and what would it take to cut that time in half?https://sled.lighthouseadvisory.consulting/streams/cisco/editions/2026-10-312026-11-02T13:21:16.063Z2026-10-31T00:00:00.000Z

Tonight’s Cisco edition looks at an AI claim inside Cisco itself. Fortune reports that Cisco’s president gave a two-person team unlimited AI tokens and a four-week proof period toward refactoring 98% of Cisco’s legacy code in 24 months. He later said this could speed moves to memory-safe languages. Independent sources push back: a modernization firm notes that the 98% is a target with no reported test method, and a preprint found automated C-to-Rust tools carried over or added memory-security bugs. Independent reporting says Cisco’s ServiceNow alliance mainly formalizes connectors that already existed. No new U.S. SLED deployment metrics surfaced.

Tonight’s Cisco edition looks at an AI claim inside Cisco itself. Fortune reports that Cisco’s president gave a two-person team unlimited AI tokens and a four-week proof period toward refactoring 98% of Cisco’s legacy code in 24 months. He later said this could speed moves to memory-safe languages. Independent sources push back: a modernization firm notes that the 98% is a target with no reported test method, and a preprint found automated C-to-Rust tools carried over or added memory-security bugs. Independent reporting says Cisco’s ServiceNow alliance mainly formalizes connectors that already existed. No new U.S. SLED deployment metrics surfaced. 5 sources · 1 cross-source patterns Cisco’s AI code-rewrite and memory-safety claims are targets; independent evidence says rewritten code must be proven on behavior and security Before crediting any vendor or in-house AI rewrite with better reliability or security, what behavior-equivalence and known-vulnerability tests will we require, and who signs off?https://sled.lighthouseadvisory.consulting/streams/cisco/editions/2026-10-302026-10-31T04:40:22.733Z2026-10-30T00:00:00.000Z

Tonight’s Cisco edition is about the security of the platforms Cisco’s AI agents will run on. On Oct 28, Lumen’s Black Lotus Labs (independent) and Cisco Talos described state-backed attacks on Cisco Catalyst SD-WAN controllers. The attackers planted malware to steal encryption keys and targeted government agencies, universities and emergency services. Cisco also says AI Defense now feeds AI security findings into ServiceNow governance, though it offers no deployment data. The Register reports that Cisco’s researchers say Project Quarry is no substitute for human oversight and has not been tested against the incident used to illustrate it. Futurum’s Oct 30 analysis sees government AI infrastructure deals but questions whether partners have the skills to deliver them. There are no new SLED deployment metrics tonight.

Tonight’s Cisco edition is about the security of the platforms Cisco’s AI agents will run on. On Oct 28, Lumen’s Black Lotus Labs (independent) and Cisco Talos described state-backed attacks on Cisco Catalyst SD-WAN controllers. The attackers planted malware to steal encryption keys and targeted government agencies, universities and emergency services. Cisco also says AI Defense now feeds AI security findings into ServiceNow governance, though it offers no deployment data. The Register reports that Cisco’s researchers say Project Quarry is no substitute for human oversight and has not been tested against the incident used to illustrate it. Futurum’s Oct 30 analysis sees government AI infrastructure deals but questions whether partners have the skills to deliver them. There are no new SLED deployment metrics tonight. 5 sources · 2 cross-source patterns The management planes Cisco wants AI agents to operate are already prime state-actor targets Before granting any AI agent or MCP server access to network management, can the agency show its controllers are patched, not exposed to the internet, and checked against published indicators? Cisco’s new AI security tools come with no published efficacy data and are framed as feeding human-run governance What local test data and which named human reviewers will validate AI security detections before they inform enforcement or compliance reporting?https://sled.lighthouseadvisory.consulting/streams/cisco/editions/2026-10-292026-10-30T03:22:17.804Z2026-10-29T00:00:00.000Z

This is the first cisco edition, and it centres on AI for running networks and security operations. At Partner Summit, Cisco launched Model Foundry and reports that a tuned network model found the right fix in 11 of 12 lab problems. It also announced MCP servers and expiring, recorded agent access to network devices, and open-sourced Project Quarry for detecting risky agent behavior. Independent coverage tempers these claims. TechTarget finds network agents remain mostly advisory, ABI says Cisco’s trust controls are asserted rather than proven, and an analyst urges buyers to contract for ownership of tuned models. The SLED deployments in view are in higher education: Internet2 AI PODs and campus SOCs. Splunk AI Assistant reached FedRAMP High.

This is the first cisco edition, and it centres on AI for running networks and security operations. At Partner Summit, Cisco launched Model Foundry and reports that a tuned network model found the right fix in 11 of 12 lab problems. It also announced MCP servers and expiring, recorded agent access to network devices, and open-sourced Project Quarry for detecting risky agent behavior. Independent coverage tempers these claims. TechTarget finds network agents remain mostly advisory, ABI says Cisco’s trust controls are asserted rather than proven, and an analyst urges buyers to contract for ownership of tuned models. The SLED deployments in view are in higher education: Internet2 AI PODs and campus SOCs. Splunk AI Assistant reached FedRAMP High. 11 sources · 3 cross-source patterns Cisco’s AI agents for network and security operations are deployed as advisors and triage aids, with humans approving fixes and containment Which network or SOC tasks will the AI agent only recommend, which will it be allowed to execute, and what recorded approval step sits between the two? Evidence for Cisco’s AI operations and trust claims is vendor-run or vendor-sponsored; independent analysts say proof and ownership terms are missing What test on our own incidents and data, and what contract language on tuned models and evaluation results, will we require before relying on Cisco AI performance or trust claims? Higher education is where Cisco AI infrastructure is most visibly deployed in SLED, but the public evidence is announcements without outcome measures If we follow these campus examples, what baseline measures (GPU queue time, troubleshooting time, alert handling) will we record before deployment so we can report results afterwards?