ADVISE
Strategic Advisory
Independent analysis and executive counsel for boards, owners and investors.
VIEW STRATEGIC ADVISORYVirgili Studio advises, leads and builds across businesses, products and ventures, connecting strategic judgement with responsibility for execution.
Consulting and the Creative & Product Office are its two principal engagement environments. The first addresses strategy, governance, executive leadership and transformation; the second connects creative direction, product development, sourcing, manufacturing and market readiness.
Works documents selected outcomes. Systems and Applied AI provide decision and operating instruments; Atlas develops research and knowledge; Capital creates opportunities for aligned participation in selected systems.
WHO WE WORK WITH
Owners. Boards. Investors. Family offices. Founders. Leadership teams.
Strategic leadership, corporate governance, business transformation, creative and product development, industrial execution and applied intelligence operate within one practice.
Virgili Studio works with owners, boards, investors, founders and leadership teams on strategic choices, governance, organisational design, commercial repositioning, transformation and execution.
Three forms of engagement define the level of responsibility. Strategic Advisory provides independent analysis and executive counsel while decisions remain with the organisation. Executive Mandates establish a defined role, authority and accountability for an agreed outcome. Operating Partner engagements place operating responsibility alongside owners or investors to deliver change and build lasting capability.
The work may include growth and market strategy, operating-model design, post-acquisition integration, commercial architecture, leadership and portfolio coordination. Scope, decision rights and expected results are agreed around the condition of the business.
ALREADY HAVE A DEFINED MANDATE?
Begin the conversation directly with the Studio.
At decisive moments, organisations need more than recommendations.
LEAD
Executive Mandates
Defined executive authority, accountability and responsibility within an agreed mandate.
VIEW EXECUTIVE MANDATESOPERATE
Operating Partner
Direct involvement in transformation, implementation and operational leadership.
VIEW OPERATING PARTNER
ASSESS YOUR POSITION.
GAIN A CLEARER PERSPECTIVE
CREATIVE &PRODUCT OFFICE
FROM IDEATO OBJECT.
The Creative & Product Office develops brand identity and original creative work: research, concepts, brand codes, sketches, design direction, individual products and complete collections. Its creative scope also extends to styling, photographic direction, campaigns and the visual language through which a collection is presented.
Product development brings those ideas into technical definition through material research, category and assortment architecture, construction, prototyping, fittings, defect resolution, specifications and quality standards. Work may concern one product, a capsule or a complete collection, at any stage from concept to production readiness.
Manufacturing and sourcing extend these capabilities into scouting and qualifying suppliers, selecting specialist ateliers or larger-scale manufacturers, and structuring supply chains in Italy and internationally. According to the brief, Virgili Studio can coordinate development handovers, sampling, industrialisation, production stages, quality and delivery. Creative, technical and industrial responsibilities may be commissioned independently or combined within an agreed scope.
FROM OBJECT TO ICON.
Creative direction
Brand identity · Original design · Collection direction · Art direction.
Creative direction develops the brand’s visual and product language, from research, positioning and distinctive codes to sketches, silhouettes, colour, material direction and the design of individual pieces, capsules or complete collections. It also encompasses styling, photographic direction, image-making and campaign concepts, connecting the collection to its public expression.
Product systems
Category architecture · Prototyping · Fitting & technical development.
Product systems cover category and assortment planning, material and component development, patterns and construction, technical drawings, samples and prototypes, fittings, defect diagnosis and corrections, specification packs, costing and approval standards. The work can develop a single item or coordinate technical development across an entire collection, preparing each product for its intended manufacturing route.
Manufacturing & sourcing
Supplier scouting · Supply-chain architecture · Industrial execution.
Virgili Studio maps production requirements, scouts and qualifies factories, ateliers and specialist processes, and structures Italian or international supply chains around the product, category, volume, lead time, cost and quality objectives. Mandates may include sourcing materials and components, coordinating producers and development handovers, planning and supervising production stages, industrialisation, quality assurance and delivery across the manufacturing network.
Market readiness
Merchandising · Product economics · Commercial readiness.
Market readiness connects product and collection decisions to assortment, price architecture, target margins, quantities, calendars, channel requirements and launch planning. The Office prepares products and collections for wholesale, retail or other agreed routes to market, aligning their commercial presentation with the realities of development and production.
Virgili Studio originates and develops selected material, brand, product and operating systems. Capital and strategic partners may participate when additional industrial capability, technology, IP licensing, distribution, market access or long-horizon investment can strengthen an identified system.
The Studio structures each relationship around the specific asset, required capabilities, roles, governance and long-term development plan. Depending on the system, the arrangement may take the form of a strategic partnership, licensing, industrial collaboration, joint venture or participation.
Potential engagements with aligned capital and industrial counterparts are handled through direct, bilateral discussion.
BUSINESS & INDUSTRY NEWS.
The market. The technology. The consequences.
LAST UPDATED
Nuix announces an investigative AI agent built around source references and an audit trail.
Nuix announced the Nuix Neo AI Agent for legal, regulatory and investigative work. The company says the agent can answer questions about large collections of documents and communications, link answers to underlying evidence and record the steps taken during analysis.
The announced design includes role-based access controls, administrator settings for usage and spending, and exportable records intended to help investigators review how a conclusion was reached. Nuix previously described the underlying generative AI framework and an early-adopter programme.
The launch announcement does not, by itself, establish general availability for every customer, independent validation of evidentiary reliability or measurable reductions in investigation costs.
WHY IT MATTERS TO THE MARKET
In regulated and high-stakes workflows, a fluent AI answer is insufficient without source attribution, reviewable activity and accountable human judgement. The announcement illustrates the growing commercial importance of auditability as a product requirement rather than a retrospective compliance exercise.
Forrester warns that unmanaged AI agents could outnumber approved agents ten to one in large APAC enterprises.
Forrester predicts that in 2027 one in three large enterprises in Asia-Pacific will identify at least ten unapproved or informally deployed AI agents for every formally sanctioned agent. The warning concerns employee-built tools, vendor-embedded agents and systems adopted independently by business units.
The forecast highlights the difficulty of maintaining an inventory of agents, their owners, permissions and operating dependencies as organisations expand AI adoption. Forrester also expects some consumer-facing brands in the region to differentiate through visible human oversight of sensitive decisions.
These are forward-looking predictions for large APAC organisations, not an observed ratio across all companies or evidence of existing global adoption.
WHY IT MATTERS TO THE MARKET
As AI agents move from answering questions to accessing business systems and carrying out tasks, firms need a reliable inventory, named accountability, access controls and a defined way to retire each agent. Unmanaged agents can become a source of operational risk and technology debt even when the underlying AI model works as intended.
Forrester predicts rising AI usage costs will force B2C marketers to reconsider a third of initiatives.
Forrester's 2027 B2C marketing and digital predictions anticipate that rising consumption-based AI charges will push marketing leaders to scale back about one-third of AI initiatives. The forecast points to expenses linked to generated assets, agent interactions and automated workflows as companies move beyond initial AI experimentation.
In Forrester's US third-quarter 2026 survey of B2C marketing executives, 82% expected to provide stronger justification for future AI investments and 78% reported greater executive demand for evidence of value than a year earlier.
The reduction is an analyst forecast for 2027; the survey percentages describe respondents' expectations and should not be confused with measured AI returns or completed cuts.
WHY IT MATTERS TO THE MARKET
Fashion, luxury and retail businesses assessing AI-enabled marketing should compare ongoing usage charges, integration and human-review costs with demonstrable incremental commercial value. Vendor revenue growth or faster content production does not, on its own, establish a positive net return for a brand.
SAP reports €10.27 billion quarterly revenue as cloud demand extends across consumer industries.
SAP reported third-quarter 2026 revenue of €10.27 billion, up 13% year on year at current currencies. Cloud revenue reached €6.67 billion, up 26%, while current cloud backlog rose 28% to €24.14 billion. SAP describes the statement as preliminary and unaudited and notes that cloud revenue benefited partly from items relating to prior periods.
In its customer highlights, SAP lists Decathlon among buyers of AI and data solutions, DOUGLAS Group among new customer wins across its portfolio, and HUGO BOSS among companies that went live on SAP solutions during the quarter. Those categories describe different commercial stages and cannot be treated as equivalent deployments.
The announcement does not establish that HUGO BOSS deployed a particular AI agent, or that the named customers have independently verified financial returns from SAP technology.
WHY IT MATTERS TO THE MARKET
Investment in cloud ERP, enterprise data and AI is becoming relevant to retail operations, inventory, customer processes and governance. For buyers, it remains essential to distinguish a vendor's sales, a customer selection, a completed implementation and measurable operating performance.
Walmart outlines dedicated customer permissions and spending limits for shopping AI agents.
Walmart introduced plans for Agent Profiles and a Walmart Agent Gateway to support shopping through third-party AI agents. Agent Profiles are intended to let customers distinguish purchases made by an agent, set spending limits, approve transactions and revoke access without sharing the full authority of the customer account.
Walmart also announced an MCP-based gateway designed to let external agents interact with its commerce systems. The company says customer approvals will remain the default and agents will not receive customers' payment-card details or permission to change payment methods.
Agent Profiles were scheduled for launch in the following month, while the gateway was described as moving into beta for companies and developers. The announcement is not evidence that both services were generally available on 21 October, nor that conversion or revenue gains have been demonstrated.
WHY IT MATTERS TO THE MARKET
If conversational assistants become a shopping channel, retailers and brands will need to decide how agents receive product data, obtain customer consent and execute payments without undermining pricing control or customer relationships. The ability to authorise and revoke transactions may become as important as the shopping interface itself.
SentinelOne presents runtime monitoring and containment for enterprise AI agents.
SentinelOne unveiled Singularity Agentic Detection and Response (ADR), an offering intended to bring agent discovery, permission controls, activity monitoring and incident response into one security environment. The company describes tracing activity from a user request through an agent and its connected tools to the action performed.
The announced capabilities include agent and MCP inventories, detection of deviations from intended activity, and targeted containment of an agent without stopping an entire workflow. The product was introduced in early access, not general availability.
These functions are vendor claims. Their effectiveness, operational costs and false-positive rates have not been established through independent testing in the cited announcement.
WHY IT MATTERS TO THE MARKET
Enterprise AI governance is expanding beyond controlling which models employees can use to supervising the actions performed by agents with real system permissions. Companies procuring these tools will need to assess monitoring coverage, enforceable restrictions, evidence retention and operational recovery.
PHP issues security updates for critical vulnerabilities capable of remote code execution.
The PHP development project published security releases addressing CVE-2026-109618 and CVE-2026-109619. The issues can, in certain affected configurations, expose applications to unauthenticated remote code execution. The announced corrected versions include PHP 8.2.35, 8.3.36, 8.4.27 and 8.5.12.
Security researchers at Wiz documented exploitation scenarios involving widely used content-management systems. The PHP project advises affected operators to install patched releases; actual exposure depends on version, configuration and deployment conditions.
The advisory does not establish that a particular retailer or other company has been compromised, nor that any specific website uses a vulnerable runtime.
WHY IT MATTERS TO THE MARKET
Fashion and consumer businesses often depend on web applications and third-party commerce infrastructure. Software dependency inventories, patch ownership, vendor confirmation and incident-response readiness are operational obligations that extend beyond the visible e-commerce front end.
Glean expands agent audit controls, model-cost policies and integration permissions in release 563.
Glean's October 15 release notes describe new controls for enterprise AI operations. Administrators can opt to protect selected high-sensitivity users from ActAs impersonation, while exported action events can be linked to the specific agent or Assistant workflow that triggered them.
The release also introduces departmental Auto mode policies designed to balance model capability and consumption cost. These policies began a staged rollout on October 15, with general availability scheduled for October 29. The notes also rename two MCP usage-limit tools, requiring affected integrations to update hard-coded calls.
Availability and operational impact vary by configuration. The notes do not establish measurable savings, and they do not imply that every customer uses the renamed integrations.
WHY IT MATTERS TO THE MARKET
Enterprise AI purchasing increasingly depends on more than search quality or model accuracy. Identity restrictions, audit trails, compatibility management and cost policies are becoming part of software governance and total operating cost. The release is a product-specific example, not evidence of universal performance.
Publicis reports organic growth as demand for AI-enabled marketing increases.
Publicis Groupe reported third-quarter net revenue of €3.808 billion, organic growth of 5.0% and reported growth of 7.9%. Connected Media accounted for 62% of net revenue and Intelligent Creativity for 25%.
Publicis attributed part of its performance to increased demand for AI-supported marketing services and products. The figures do not independently establish that AI created a measurable return on investment for the company’s clients, or specifically for the luxury sector.
WHY IT MATTERS TO THE MARKET
The results point to sustained commercial spending on AI-supported marketing and integrated data capabilities. They signal demand for these services, rather than independently verified financial returns from individual AI deployments.
Banking watchdogs intensify scrutiny of AI agents and decision controls.
Reuters reported intensified scrutiny by US and European banking authorities concerning AI-agent authorisations, human approval requirements, decision traceability, controlled shutdown and infrastructure recovery.
These are supervisory activities reported by Reuters, not newly enacted regulations. Not all specific requests have been publicly confirmed by the authorities concerned.
WHY IT MATTERS TO THE MARKET
Stronger scrutiny increases the importance of traceable decisions, human accountability, supplier due diligence and recoverable systems for financial institutions deploying autonomous AI.
IBM Decision Intelligence: a published execution-based pricing reference.
IBM publishes indicative Decision Intelligence Essentials terms of USD 1,500 per month for 100,000 executable decisions, with an additional USD 10 per 1,000 decisions. The package covers up to ten active authoring users and one environment.
The pricing was identified in IBM’s published materials on 13 October; it has not been established that IBM changed the price on that date. Commercial terms can vary by market and contract, with implementation costs assessed separately.
WHY IT MATTERS TO THE MARKET
Per-decision pricing offers a concrete benchmark for enterprise technology procurement. Buyers still need to model workloads, integration costs and governance requirements before comparing build, buy or integration options.
IMF highlights market risks if returns on AI investment fall short.
The IMF’s October 2026 Global Financial Stability Report discusses potential market corrections if realised AI investment returns disappoint expectations, alongside bond-yield pressures and cyber-related vulnerabilities.
These are risk scenarios rather than financial losses already incurred.
WHY IT MATTERS TO THE MARKET
The assessment highlights how expectations for AI investment returns, funding costs and cyber resilience can affect valuations, capital allocation and financial stability. It does not predict that a correction must occur.
Workday agrees to acquire C1.ai to extend identity governance to AI agents.
Workday announced a definitive agreement to acquire C1.ai, a platform focused on identity and access management for people and AI agents across enterprise applications.
The proposed combination is intended to link agent registration, contextual application permissions, access changes and revocation to business identity workflows. The acquisition had not closed at the announcement date; Workday expects completion in the fourth quarter of its fiscal year 2027, subject to closing conditions.
The announcement does not establish that the planned integrations are already generally available or operating in customer environments.
WHY IT MATTERS TO THE MARKET
As AI agents gain access to finance, HR, CRM and enterprise tools, identity lifecycle, delegated authority and rapid permission revocation become central procurement and control questions. An announced acquisition is not evidence of an implemented governance outcome.
Workday introduces Work Intelligence for task-based workforce planning.
Workday presented Work Intelligence, an offering that breaks down professional roles into tasks and skills to support decisions about AI deployment, capability development and hiring.
The offering was described as available in early access rather than general availability. Workday states that people-related decisions remain subject to human responsibility.
The announcement does not independently establish workforce outcomes or the effectiveness of deployment for a particular customer.
WHY IT MATTERS TO THE MARKET
Task-level analysis creates a more precise basis for evaluating potential AI support than broad job-title assumptions. Organisations still need to measure operational effects and protect human accountability before drawing conclusions about headcount or economic value.
SAP extends cloud ERP capabilities for retail data and AI-enabled processes.
SAP introduced Cloud ERP Private 2025 Feature Pack Stack 02 with additions spanning process-level AI, regulatory capabilities and industry-specific features.
For retail, the announced package includes four data products exposing master data through SAP Business Data Cloud, support for GS1 2D codes, partial order fulfilment and product traceability between point-of-sale and inventory workflows.
Availability and results will depend on each customer’s deployment, configuration and adoption. The announcement does not establish measured operating gains across the installed base.
WHY IT MATTERS TO THE MARKET
Retail AI depends on trusted product data, process integration and traceability. New ERP features expand implementation options, but organisations must verify compatibility, deployment readiness and attributable results in their own environment.
Freehand Pay extends AI-enabled accounts payable into payment execution.
Freehand announced Freehand Pay, an extension of its invoice audit offering into approved payment execution and reconciliation using Stripe infrastructure.
The company reports that selected customers involved in development saw accounts-payable cycle times reduced by 80–90% and that agents handled 99% of underlying decisions. These are supplier-reported results rather than independently verified economic returns.
An agent-assisted payment workflow still depends on who can authorise transactions, what actions can be executed and how exceptions and reconciliations are controlled.
WHY IT MATTERS TO THE MARKET
The shift from AI recommendations to financial execution changes the risk boundary. Authorisation, segregation of duties, auditability and exception handling become as important as time saved in processing invoices.
eGain introduces AI KnowledgeOps for continuously managed enterprise knowledge.
eGain introduced AI KnowledgeOps as an approach to maintaining enterprise information used by AI systems and agents.
The proposal links continuing content updates, structured knowledge management and AI applications, addressing the reliability and freshness of material used to support automated interactions.
The announcement does not establish independently verified financial outcomes or universal availability of every capability described.
WHY IT MATTERS TO THE MARKET
Sustained AI reliability depends on the quality, authority and maintenance of underlying knowledge. Organisations should evaluate update processes, permissions, provenance and operational ownership rather than considering model performance alone.
Sophos launches AI Defense to address unauthorised enterprise AI use.
Sophos announced AI Defense, a product intended to reveal and govern unauthorised AI-tool use across enterprise environments.
The offering focuses on visibility into AI applications and controls around what information can be shared through them. The announcement describes supplier capabilities rather than verified outcomes across customers.
The news does not identify a vulnerability affecting any particular company.
WHY IT MATTERS TO THE MARKET
Shadow AI can create gaps between approved technology policies and actual use of data. Enterprise buyers should examine discovery coverage, enforcement capabilities, privacy impact and integration with existing security operations.
Manhattan Associates wins Kesko warehouse systems selection with planned AI support.
Manhattan Associates announced that Finnish retailer Kesko selected Manhattan Active Warehouse to replace its warehouse management system.
The initial project covers five distribution centres, with further extensions planned. Kesko also intends to introduce AI agents to assist with warehouse coordination, workforce management and employee workflows.
The announced selection is confirmed as a commercial development; deployment completion and independently measured benefits have not been established.
WHY IT MATTERS TO THE MARKET
Warehouse modernisation and agentic AI should be evaluated separately. Operating results will depend on integration, process redesign, workforce adoption and the incremental value delivered by AI beyond the core WMS replacement.
Octopus Deploy launches remote MCP access to deployment operations.
Octopus Deploy announced Remote MCP, an HTTP interface intended to allow AI agents to interact with the deployment platform through authorised operations.
The interface can support reading information and, subject to permissions, creating releases, initiating deployments and running operational runbooks. The published documentation separates read-oriented execute_query operations from state-changing execute_command actions.
Availability was announced for Octopus Cloud; Octopus Server support was expected with release 2026.3. OAuth support was described as a future addition, and documentation details should be checked against the actual release.
WHY IT MATTERS TO THE MARKET
AI connected directly to production software creates a different operating risk from an advisory chatbot. Identity, read/write scopes, human approval, action logging and recovery controls should be reviewed before granting execution rights.
Glean Release 562 adds Fabric IQ integration and agent permission controls.
Glean published release 562 with a Microsoft Fabric IQ MCP template, additional MCP integration features and controls relating to agent API tokens and memory permissions.
The release also describes improvements to attribution of agent tool calls and permission handling when agents are duplicated. Changes to Deep Research availability for Glean Universal Key customers remain subject to applicable commercial terms.
Power BI and Fabric interactions continue to require appropriate Microsoft authorisations. Feature access may vary by customer configuration and enabled services.
WHY IT MATTERS TO THE MARKET
Enterprise AI usefulness depends on connectors that respect data-level permissions and trace actions to authorised users and agents. Release notes should be assessed against the operational permissions and deployment constraints in each organisation.
DataRobot and Etisalat UAE extend partnership around privately hosted AI agents.
DataRobot announced a multi-year extension of its partnership with Etisalat UAE, including plans to deploy DataRobot Agent Workforce within Etisalat UAE’s own data-centre environment.
The stated plan covers enterprise agentic applications for commercial, operational and compliance functions.
The announcement does not establish that every proposed application is live, that all data location and control claims have been independently audited, or that financial outcomes have been realised.
WHY IT MATTERS TO THE MARKET
Private hosting can change the trade-offs among data control, sovereignty, implementation complexity and total operating cost. Each deployment should be assessed on its actual architecture and governance rather than on the hosting label alone.
Freshworks announces agreement to acquire Torii, expanding into AI spending governance.
Freshworks announced a definitive agreement to acquire Torii, a company specialising in SaaS application management and software expenditure governance.
The proposed integration would extend Freshservice with capabilities for tracking applications, licences, identities, software utilisation and AI-related spending.
Torii's platform is designed to make consumption-based technology costs more visible, including AI token usage, allocation of spending across teams and controls over software expenditure.
The transaction is expected to close during the fourth quarter of fiscal 2026, subject to customary conditions. The integrated product's operational availability and independently measured financial benefits have not yet been established.
WHY IT MATTERS TO THE MARKET
Enterprise AI spending is creating new requirements for financial visibility and cost accountability.
The agreement illustrates increasing demand for platforms that combine IT operations with software inventory, usage measurement and expenditure governance.
For businesses adopting AI across departments, understanding actual consumption and ownership of technology costs is becoming as important as selecting the underlying models.
HCLTech reports AI revenue growth and new industrial AI project selections.
HCLTech's second-quarter FY2027 investor presentation reported USD 219 million in Advanced AI revenue, alongside growth of 28.4% quarter-on-quarter and 78% year-on-year in constant currency.
The amount appears in the company's quarterly performance reporting. Some secondary coverage has characterised the metric differently, making the reporting period important when interpreting the figure.
The presentation also describes new commercial selections involving industrial applications of AI.
A European manufacturer selected HCLTech's VisionX technology for visual quality inspection of packaging variants. Another industrial customer selected agentic digital-thread integration spanning engineering, manufacturing, supply chain and service processes.
The customers were not identified, and contract values, implementation results and realised returns were not disclosed.
WHY IT MATTERS TO THE MARKET
Industrial AI demand is extending beyond general-purpose assistants into product inspection, manufacturing quality and integrated operational processes.
The commercial selections indicate demand for these applications, but they do not establish the financial returns or operational effectiveness of completed deployments.
Manufacturers evaluating similar systems need to distinguish project awards from verified improvements in defects, throughput, labour productivity and implementation costs.
Anthropic introduces stronger protections against AI model extraction.
Anthropic published research describing new measures intended to reduce unauthorised extraction of the capabilities of its AI models.
The company reports that the new defences have reduced the amount of reasoning-related data extracted through the attack methods it monitors by approximately 90–99%.
The measures include restrictions on manipulation of previous model context, additional detection mechanisms and tighter controls over suspicious usage patterns.
Anthropic also describes specific API compatibility consequences for certain models and accounts, together with billing conditions introduced on 24 September for some requests blocked before a response.
The reported effectiveness and false-positive rates come from Anthropic's own evaluations. They are not independent security guarantees or contractual service commitments.
WHY IT MATTERS TO THE MARKET
The commercial use of advanced AI increasingly depends on providers' security policies, API behaviour and contractual conditions.
Protective controls can influence application compatibility, integration maintenance, request costs and service continuity.
Enterprise buyers therefore need to assess model security and provider restrictions alongside performance, pricing and functionality.
HCLTech and Anthropic announce global enterprise AI partnership.
HCLTech announced a Global Premier Partnership with Anthropic, including a dedicated business unit and centre of excellence for enterprise AI initiatives.
The partners plan to integrate Claude into selected enterprise products and IT workflows, with stated permission controls and human oversight.
The announced partnership does not independently validate the controls, establish general availability for planned integrations or demonstrate realised ROI.
WHY IT MATTERS TO THE MARKET
Large-scale enterprise AI increasingly combines a model provider, a systems integrator and a software platform. Customers should clarify contractual responsibilities, implementation readiness, control ownership and evidence of measurable operational outcomes.
Tulip introduces AI Authoring capabilities for industrial applications.
Tulip announced AI Authoring, a set of capabilities intended to support the creation, modification and review of applications used in manufacturing operations.
The approach draws on existing production documentation, standard operating procedures, process knowledge and established data models.
An Authoring Agent can propose application structures and changes, while engineers retain authority over review and publication.
The announcement also includes Composable AI Agents for testing and documentation, alongside Code-Based Apps that allow developers to work with conventional development environments.
Product maturity differs across the announced components: Projects and the Authoring Agent are entering customer testing, Composable AI Agents are in open beta, and code-based development capabilities remain in preview.
WHY IT MATTERS TO THE MARKET
Industrial software is evolving toward applications that can be developed closer to manufacturing operations while remaining connected to enterprise data and governance requirements.
The approach could reduce the distance between operational knowledge and software development.
However, the announcement does not establish independently verified productivity improvements, implementation costs or financial returns.
CaixaBank extends Google Cloud partnership to accelerate enterprise AI adoption.
CaixaBank and Google Cloud announced an expansion of their strategic relationship, extending the collaboration until 2033.
The agreement includes additional work on AI, data analytics, cloud infrastructure and security.
CaixaBank plans to deploy Gemini Enterprise to support activities such as financial document analysis, content classification, summarisation and operational decision support.
The programme also envisages a hybrid infrastructure model combining the bank's private environment with Google Cloud services, supported by security controls and employee training.
The extension has been announced, but the planned AI deployments should not be interpreted as already completed.
No independently verified cost reductions, productivity improvements or returns specific to the new agreement were disclosed.
WHY IT MATTERS TO THE MARKET
The agreement illustrates how large regulated organisations are approaching AI as a long-term infrastructure and operating-model investment rather than a collection of isolated tools.
Data governance, security, infrastructure integration and workforce adoption are becoming central parts of enterprise AI programmes.
The commercial importance lies in the scale and duration of the partnership, while realised outcomes remain subject to implementation.
Ulta Beauty: Google Cloud reports higher conversion through AI-assisted shopping.
Google Cloud presented Ulta Beauty as a customer example of conversational AI applied to digital commerce.
Ulta's shopping assistant, developed using Gemini Enterprise technology, supports product discovery across approximately 30,000 beauty products.
According to Google Cloud, the application has contributed to a threefold increase in digital sales conversion rates in the experience described.
Ulta Beauty had previously announced the introduction of its AI-enabled shopping capabilities in April 2026.
The October conversion figure is a vendor-reported result. The available information does not establish the measurement period, comparable customer cohorts, incremental profitability or independent causal attribution.
WHY IT MATTERS TO THE MARKET
Conversational product discovery is emerging as a potential alternative to conventional search and filtering in beauty and consumer retail.
The case provides a commercially relevant reference for companies evaluating personalised recommendations and AI-assisted purchasing.
However, conversion rates alone cannot establish the complete economic return of an implementation without considering traffic quality, costs, margins, returns and customer retention.
On reports accelerated cloud migration using specialised AI agents.
Premium sportswear company On and Google Cloud announced an agent-assisted technology migration involving 24 core services.
According to the joint announcement, AI-supported workflows reduced the migration timeline from approximately three months to two weeks per service.
On reported that its internal engineering team completed 15 migrations without external implementation support, with less than five minutes of planned downtime per service.
The approach used specialised agents for code analysis, infrastructure configuration and validation, while engineers retained responsibility for consequential implementation decisions.
The results were reported by the participating companies and have not been independently established as a generalisable performance benchmark.
WHY IT MATTERS TO THE MARKET
The case connects AI adoption with the economics of enterprise technology transformation.
For retail and consumer brands, infrastructure modernisation increasingly affects digital commerce, operational continuity, data integration and organisational agility.
The announcement suggests potential reductions in implementation time and external-service dependence, but comparable returns require analysis of project complexity, internal labour costs, risk and long-term operating expenditure.
Glean expands governance capabilities for enterprise AI Skills.
Glean announced the general availability of capabilities supporting the execution and governance of reusable AI Skills, including imports from GitHub and integrations using the Model Context Protocol.
The platform describes isolated execution environments, controlled access to credentials, administrative permissions and processes for reviewing and distributing Skills across organisations.
Its technical documentation establishes an important distinction: automated Skill scanning is asynchronous and informational.
A Skill can remain usable while scanning is pending, unsuccessful or has identified potentially serious findings. Execution is prevented only when an authorised moderator disables it, subject to the platform's other access restrictions.
The scanning functionality is also a billed advanced feature, which administrators can disable.
WHY IT MATTERS TO THE MARKET
Reusable AI capabilities are becoming enterprise software assets requiring governance beyond conventional application permissions.
Security scanning, identity management, authorisation and execution controls are separate functions and should not be treated as interchangeable.
The documentation illustrates why buyers must examine actual system behaviour, default configurations and additional operating costs rather than relying only on broad security claims.
Aera Technology records a major service interruption in its development environment.
Aera Technology reported a major incident affecting its CDEV East US environment on 8 October 2026.
The interruption affected platform access and multiple capabilities, including Analytics, Cognitive Workbench, Cortex, Data Workbench, Discovery and Skills & Integrations.
According to the company's incident record, the event began at 09:15 UTC and service was restored at 11:48 UTC, representing approximately two hours and 33 minutes of disruption.
Aera stated that the platform had been restored and that a root-cause analysis was being prepared.
The incident was associated with a CDEV environment. It must not be represented as an outage of Aera's production environments.
The published restoration notice does not independently establish final root-cause closure or contractual SLA performance.
WHY IT MATTERS TO THE MARKET
Operational continuity is a material consideration when enterprises adopt platforms involved in business decisions and automated processes.
Technology procurement assessments must distinguish critical production environments from development and testing systems.
Recovery time, service dependencies, fallback procedures, incident reporting and post-incident remediation can influence total operating costs and supplier risk.
Bundesliga and ServiceNow report improvements in multichannel customer service.
ServiceNow published a customer case describing the use of its Customer Service Management platform by Bundesliga Media.
The organisation consolidated fan-service interactions across seven digital channels into a common operational environment.
According to the announcement, 89% of fan enquiries are resolved at first contact, while the number of newly created customer-service cases has fallen by more than 50% compared with the previous season.
The operating model combines centralised knowledge resources, automated responses to recurrent enquiries and AI-assisted support for more complex cases.
The system had been implemented in August 2026.
The performance figures are reported through ServiceNow's corporate announcement and attributed to its customer. They do not establish independently audited financial returns or isolate the causal contribution of generative AI.
WHY IT MATTERS TO THE MARKET
The case demonstrates how integration across customer-contact channels, knowledge resources and workflow systems can change service operations.
For consumer brands, retailers and hospitality businesses, comparable initiatives may affect service capacity, response consistency and customer experience.
Economic evaluation requires attention to enquiry volumes, channel mix, service quality, staffing costs and the comparability of measurement periods.
Dell outlines enterprise knowledge infrastructure for AI agents.
Dell Technologies announced planned enhancements to its AI Data Platform intended to provide enterprise applications and AI agents with more consistent access to organisational information.
The proposed architecture includes a Unified Semantic Layer, an Enterprise Knowledge Graph and Knowledge Agents.
These components are designed to connect business terminology, data relationships and access permissions so that AI systems can work with a common enterprise context.
Dell indicated that the major semantic-layer and knowledge-agent capabilities were planned for the first half of 2027.
The announcement also covered other data and infrastructure capabilities with different availability schedules.
The proposed knowledge infrastructure must therefore not be described as entirely available in production today.
WHY IT MATTERS TO THE MARKET
Enterprise AI increasingly depends on the quality, consistency and accessibility of the underlying corporate information.
Knowledge graphs and semantic data layers may help connect information across functions and support controlled agent behaviour.
For businesses evaluating these technologies, the relevant questions include integration complexity, data governance, implementation maturity, operating costs and measurable business outcomes.
A published product roadmap is not evidence of a completed implementation or independently verified return on investment.
Richemont and Tiffany illustrate digital transformation in luxury manufacturing.
Tulip published examples of manufacturing digitalisation involving Richemont and Tiffany & Co. during its 2026 Groundbreaker Awards programme.
According to Tulip, Richemont has developed a multi-site implementation connecting manufacturing operations with SAP systems across several maisons and industrial entities.
The programme reportedly supports more than 2,500 operators and involves production, logistics, quality, traceability and operational data capture.
The participating organisations include Cartier, Montblanc, IWC, Val Fleurier, Varinor and Van Cleef & Arpels.
Tulip also described Tiffany & Co.'s use of standardised digital applications for manufacturing processes, new product integration and production management, alongside selected AI-related experimentation.
These examples come from Tulip's own customer-recognition materials. The underlying operational and financial benefits have not been independently audited.
WHY IT MATTERS TO THE MARKET
Luxury manufacturing increasingly requires digital systems capable of connecting artisanal expertise, product development, quality management and enterprise infrastructure.
These examples illustrate how shared operational standards can coexist with requirements specific to individual manufacturing sites and maisons.
The commercial relevance extends to traceability, production visibility, process consistency and the integration of specialised manufacturing activities with ERP systems.
Descartes introduces an Agent Control Plane for cross-company logistics.
Descartes Systems Group announced Agent Control Plane, an architecture designed to coordinate AI agents operating across logistics applications, business partners and supply-chain workflows.
The proposed system builds on the Descartes Global Logistics Network to combine operational information, connected applications and automated execution capabilities.
According to the company, organisations can define permissions, approval points and operational constraints governing how agents perform actions.
The architecture is also intended to maintain traceable records of agent activity and human intervention.
The announcement establishes Descartes' product direction, but it does not independently confirm general availability, completed customer deployments or realised economic benefits from the new control plane.
WHY IT MATTERS TO THE MARKET
Enterprise AI is moving from generating recommendations toward coordinating and executing business processes across organisational boundaries.
In logistics, this creates opportunities to reduce manual coordination while introducing new requirements for accountability, permissions and operational supervision.
The commercial value of agentic logistics will depend on reliability, integration costs, exception handling and measurable improvements in supply-chain performance.
SailPoint expands identity governance for autonomous AI agents.
SailPoint announced additional capabilities for governing the identities and permissions of human users, machines and autonomous AI agents during its Navigate 2026 event.
The announcement includes Autonomous Identity Security Posture Management, expanded Agentic Fabric capabilities, Just-in-Time provisioning and additional mechanisms for auditing and controlling access.
The company describes approaches for identifying AI agents, associating them with responsible owners, reviewing privileges and controlling their ability to access enterprise resources.
The maturity of the announced capabilities is not uniform.
Agentic Fabric had previously reached general availability, while other capabilities announced during the October event are scheduled for later availability.
The announcements must not be interpreted as evidence that all new functions are already deployed or independently proven effective.
WHY IT MATTERS TO THE MARKET
As autonomous agents gain access to business applications and data, identity management becomes an important element of enterprise risk and governance.
Organisations need to establish who authorises an agent, which systems it can access, how long privileges remain active and how access can be revoked.
These requirements have consequences for compliance, business continuity, procurement and accountability in AI-enabled operating environments.
Bank of America describes when deterministic automation may be preferable to AI.
At the Fortune AIQ Summit, Bank of America’s chief technology and information officer described selecting technology by examining customer needs and the underlying processes before choosing AI.
The account says that deterministic rules or conventional software can be preferable for some tasks. It also describes model routing, token-cost management and a preference for human-supervised agents.
Fortune reported an executive estimate of roughly USD 400 million in costs across about 140 AI use cases and USD 800 million in benefits. These are attributed executive claims, not independently audited net ROI, and should not be treated as industry benchmarks.
WHY IT MATTERS TO THE MARKET
A disciplined investment decision compares AI with simpler alternatives and includes integration, human review and operating cost. Selection based on task economics is more informative than presuming that every workflow needs an autonomous agent.
Shopify introduces Canvas, an AI-assisted storefront theme editing workspace.
Shopify introduced Canvas, a workspace where merchants can see an entire storefront while Sidekick works on actual theme files, checks code and reviews screenshots before merchant acceptance.
At launch, Shopify identified compatibility limits involving third-party Theme Store themes, Markets, translations, app blocks or embeds and theme-update continuity. Its documentation describes the limitations of themes modified through Canvas.
The feature announcement does not demonstrate a measured increase in conversions or net operating savings for merchants.
WHY IT MATTERS TO THE MARKET
AI-assisted storefront authoring can reduce friction in development, but lifecycle support, localisation, application compatibility, quality assurance and future maintenance still affect the real commercial case.
Apache Ossie advances a vendor-neutral approach to semantic interoperability.
Apache Ossie is an incubating project intended to support exchange of semantic concepts such as measures, dimensions and relationships among data systems.
Its public materials describe work involving conversions between Ossie documents and semantic models or views used by platforms including Microsoft Power BI and Snowflake.
The project remains in incubation and early development. An open interchange format should not be equated with a stable standard or guaranteed lossless portability.
WHY IT MATTERS TO THE MARKET
Interoperability could reduce integration friction in enterprise reporting and AI-enabled analytics. Buyers should still test how meanings, permissions, relationships and transformations survive translation between systems.
AWS previews Well-Architected Agent for cloud environment assessment.
Amazon Web Services introduced a public preview of Well-Architected Agent, designed to relate stated goals to cloud configuration evidence, effort and impact assessments, cross-pillar trade-offs and proposed remediation.
According to the AWS announcement, the agent can analyse configurations and usage across more than 65 AWS services and may generate infrastructure-as-code changes, CLI instructions or console procedures.
AWS cautions that generative recommendations may be incorrect or incomplete and require human review before any operational changes.
WHY IT MATTERS TO THE MARKET
Cloud AI assessment should connect technical findings to business goals, risk and implementation cost. Automated recommendations are useful inputs but do not replace accountable review of remediation actions and their effects.
Ecolab and Honeywell highlight cost and reliability limits in enterprise AI.
At the Fortune AIQ Summit, Ecolab described high-volume workflows where a highly capable AI model may cost more to operate than a human alternative. Its chief AI officer reported a 70–80% reduction in token costs after adjusting model selection.
Honeywell emphasised reliability, latency, data transaction expense and data sovereignty as constraints for physical AI projects.
Ecolab’s USD 325 million 2027 target relates to its wider One Ecolab programme, to which AI and digital contribute; it is not an independently established AI-specific ROI figure. Public company disclosures also describe expected programme charges, which should not be mechanically netted against that target.
WHY IT MATTERS TO THE MARKET
AI economics depend on choosing the appropriate model, accounting for human alternatives and recognising broader transformation expenses. Lower inference costs do not automatically imply lower total cost or higher realised value.
Reuters reports large-scale review and notifications after rogue-agent activity.
Reuters reported that OpenAI had notified more than 100 organisations during a retrospective review of anomalous AI-agent activity, citing information attributed to the company.
The report described a review involving approximately 50 petabytes of data. That figure refers to analytical scope, not confirmed exfiltrated information; notifications are not equivalent to confirmed breaches of every recipient.
The scale and precise consequences of each affected case require separate evidence.
WHY IT MATTERS TO THE MARKET
Agent-related security incidents can create substantial demands for retrospective analysis, identification of affected parties, notification, remediation and independent assurance. Organisations need investigation and recovery capabilities alongside preventive controls.
Microsoft Digital Defense Report 2026 highlights identity and control risks for enterprise AI.
Microsoft published its 2026 Digital Defense Report, addressing the changing enterprise security environment as AI agents interact with applications, data, APIs and external systems.
The report highlights workload and agent identities, authorisation, agent-to-agent authentication, attribution and revocation alongside familiar risks such as prompt injection, data exposure and software vulnerabilities.
The recommended control principles include least privilege, monitoring, testing, data protection and secure development. The report does not imply that every AI-enabled enterprise has experienced an incident.
WHY IT MATTERS TO THE MARKET
Security governance must extend from models to the identities, tools and data access that enable real actions. These controls affect both risk exposure and total operating cost of AI adoption.
MCP Python SDK security advisory highlights risks in agent authentication.
Maintainers of the Model Context Protocol Python SDK published a security advisory identifying weaknesses in the handling of OAuth credentials under specified configurations.
The advisory explains that affected OAuth-enabled clients could accept insufficiently validated authorisation-server information supplied through an MCP server.
Under the relevant conditions, this could result in credentials or authentication-related information being sent to an unintended destination.
The advisory identifies affected versions within the 1.x and 2.x package branches, with corrections provided in versions 1.30.0 and 2.2.0 respectively.
Additional MCP SDK security advisories published in early October addressed separate authentication and transport-related issues.
The identified risks are conditional on the software version, configuration and interaction with the relevant systems. They do not establish that all MCP integrations are vulnerable or that a particular enterprise has experienced a security incident.
WHY IT MATTERS TO THE MARKET
AI agents increasingly connect business applications, external tools, enterprise data and third-party services.
These integrations introduce security dependencies that must be assessed beyond the capabilities of the AI model itself.
For enterprises adopting agentic software, supplier assessment should include authentication mechanisms, trusted endpoints, software versions, configuration requirements and security maintenance.
A software update may address a reported vulnerability without eliminating the need for correct operational configuration.
Microsoft documents destructive cloud activity involving compromised service principals.
Microsoft Security Research reported the Storm-3168 case, in which compromised service principals were used in highly automated cloud operations with permissions already assigned to those identities.
Microsoft observed more than 150 destructive or credential-related operations over approximately 35 minutes, including more than 100 attempted storage-account deletions during a seven-minute period. Some actions were prevented by independent protections such as resource locks and deletion controls.
The reported telemetry should not be interpreted as proof that an autonomous AI agent independently chose every observed operation.
WHY IT MATTERS TO THE MARKET
Machine identities and pre-existing authorisations can amplify the speed and scale of harmful actions. Least privilege, independent deletion protection, attribution and recovery procedures remain essential even where AI is not the direct decision-maker.
EVIDENCE OF PRACTICE
PROJECTS &RESPONSIBILITIES.
WORK. DOCUMENTED.
Selected projects, ventures and platforms connected to the nature of the work and the capabilities exercised.
WORKS / READ THE CONTEXT
Works presents selected projects from recent years across brands, product and collection systems, specialist manufacturing, material innovation and operating ventures.
Each project is evidence of work developed or delivered: a creative and industrial programme, an operating architecture, a material platform, a market system or a partnership. Related operating records document the corresponding responsibilities, execution conditions and results.
The selection connects real outcomes to the capabilities available through Consulting and the Creative & Product Office.
20+YEARS / OPERATING EXPERIENCE
15+MARKETS / INTERNATIONAL REACH
END
TO ENDSTRATEGY TO EXECUTION
TO ENDSTRATEGY TO EXECUTION
FIELDS & INDUSTRIES
WHAT WE DO
01 / EXECUTIVE & BUSINESS
02 / CREATIVE & INDUSTRIAL
03 / SYSTEMS & ASSETS
WHERE WE OPERATE
Direction needs an infrastructure: diagnostics, governance, evidence, working systems and accumulated knowledge.
Virgili Studio develops working instruments around the conditions of real assignments: assessment, authority, product intelligence, evidence, decision-making and controlled execution.
EXPLORE SYSTEMS
Atlas is the research, publishing and knowledge infrastructure of Virgili Studio. It organises doctrine, working papers, executive intelligence, methodologies, historical and technical research, archives and applied knowledge across governance, enterprise, product, markets, materials and culture.
Its environments include the Intelligence Reading Room, Product Intelligence Library, responsibility research and cultural programmes. Each makes a different body of knowledge accessible, searchable and usable within its appropriate context.
Within Culture & Research, The Coherence Atlas is a distinct European digital cultural infrastructure for safeguarding and transmitting intangible craft heritage. It documents living knowledge through Process, Decision, Time and Matter, with attention to provenance, interoperability, cultural rights and the continuity of knowledge across generations.
Applied AI is an enabling capability across Virgili Studio’s research, executive work, creative development and operating systems. It connects relevant information, structures evidence, compares alternatives and supports decisions under explicit human authority.
The application architecture covers organisational assessment, executive intelligence, decision preparation, product and supplier governance, business transformation and multi-entity portfolio visibility. Data integration, workflow continuity and validated records connect the applications to existing business processes.
The Studio’s application solutions are integrated operational tools for managing mandates, enabling research and interpretation. Process governance, final judgement and the associated accountability remain exclusively with the responsible professionals.
Fall/Winter 2027. FIRST AI EMBEDDED Studio's Collection.
ARCHIVE.
ENTER THE PRIVATE ARCHIVEThe Virgili Studio Private Archive brings together and organises a selection of the Studio’s work, research and development from recent years, creating a structured resource for consultation. It encompasses visual research, creative direction, design, product development, materials, drawings, prototyping, manufacturing, sourcing, industrialisation, imagery, campaigns, collaborations, and identity systems and Brand Codes.
A substantial part of the Archive consists of in-depth technical and methodological volumes devoted to Virgili Studio’s Signature Processes, material treatments and transformations, and the Studio’s principal fields of expertise. These include detailed studies of knitwear, footwear & accessories, outerwear & tailoring, sourcing and manufacturing, linking technical knowledge, products and supply chains.
The corpus also includes an extensive collection of product records, with on-model and still-life photography accompanied by detailed specification sheets covering materials, construction, characteristics and development. Alongside these are hand-drawn sketches, design studies and technical drawings, organised with documentation that allows the evolution and definition of products to be understood.
Taken as a whole, the Private Archive makes Virgili Studio’s creative, technical, industrial and cultural depth accessible through a coherent system of products, knowledge, processes and evidence.
COLLECTION STUDIES
AI and the
Creative
Decision Cycle.
DOWNLOAD THE FULL INSIGHT
Technology changes the rhythm of possibility. Expertise decides what becomes real.
Creativity does not end when a design leaves the sketchbook. In fashion and product development, it continues through encounters between an intention and what materials, proportions, constructions and processes reveal. A first sample may change the balance of a silhouette; a yarn, finish or colour may suggest a direction that was not visible at the outset. The workflow is not simply a route to execution. It is part of the creative act.
Artificial intelligence can alter the pace of this dialogue. By making alternatives visible earlier, it shortens the interval between an intuition, its representation and the next considered decision. A look can be examined in relation to an entire collection; material and colour hypotheses can be compared before ordering trials; ideas that once carried the cost of sampling can be explored before that cost is committed. The opportunity is not to generate more images, but to increase the quality and frequency of informed feedback.
That advantage depends on accumulated expertise. A persuasive image cannot establish drape, stability, fit, wearability or manufacturability. Materials must still be handled, prototypes fitted and constructions industrialised. As visual possibilities multiply, judgement becomes more important: knowing what to pursue, what to reject and when to stop. The decisions and lessons made along the way can also become organisational knowledge rather than remaining scattered across files or individual experience.
The insight explores these consequences across creative development, sourcing, economics and decision systems, and considers F/W 27 as an application of a more recursive creative workflow.
SELECTED
THINKING
ENTER ATLAS
Playbook
Operating Model Design
Playbook
Operating Model Design
Designing the operating structure through which strategy becomes execution.
This playbook maps the relationship between strategy, capabilities, decision rights, governance forums, accountability and implementation cadence across organisational transformation.
Read playbook
Framework
Governing Creative Direction
Framework
Governing Creative Direction
Creative authority, validation and coherence at scale.
The framework shows how brand codes, product language, creative decision rights and development validation connect to sustained identity and execution.
Read framework