The second WP Agency Forum took place on 7 October 2026 at Talent Garden Madrid, drew 150 in-person attendees and over 200 online, and was organized by SiteGround España and Modular DS to focus on business strategy amid rapid AI-driven change in the sector.
At a three‑day INMA Newsroom and AI Master Class (20 October 2026), speakers described a shift in which newsrooms treat AI not only as a tool but as a means to redesign production, prioritise higher‑value journalism, and develop new revenue and audience strategies.
At a three‑day INMA Newsroom and AI Master Class (20 October 2026), speakers described a shift in which newsrooms treat AI not only as a tool but as a means to redesign production, prioritise higher‑value journalism, and develop new revenue and audience strategies.
At a 20 October 2026 INMA study tour stop at Salesforce Tower, Salesforce executives framed AI as a force for organisational coordination, outlined a four-layer agentic operating model, warned about costs of replacing CRM with general models, and demonstrated an agent-to-agent ad-sales pilot involving Nine and Dentsu that remains in pilot.
At a 20 October 2026 INMA study tour stop at Salesforce Tower, Salesforce executives framed AI as a force for organisational coordination, outlined a four-layer agentic operating model, warned about costs of replacing CRM with general models, and demonstrated an agent-to-agent ad-sales pilot involving Nine and Dentsu that remains in pilot.
Independent benchmarks show agents reliably complete short, well-specified tasks but fail as tasks grow longer, more interactive or underspecified; reported success rates include 30% (TheAgentCompany, September 2025), about 58% when all information is given vs 35% when information arrives piecemeal (Salesforce CRMArena-Pro, May 2025), and measured median successful-task horizons of roughly 50 minutes (March 2025) rising to about 2 hours 17 minutes for GPT-5 in later METR measures.
Independent benchmarks show agents reliably complete short, well-specified tasks but fail as tasks grow longer, more interactive or underspecified; reported success rates include 30% (TheAgentCompany, September 2025), about 58% when all information is given vs 35% when information arrives piecemeal (Salesforce CRMArena-Pro, May 2025), and measured median successful-task horizons of roughly 50 minutes (March 2025) rising to about 2 hours 17 minutes for GPT-5 in later METR measures. Industry signals echo these limits: Gartner forecasted on 25 June 2025 that over 40% of agent projects would be abandoned by end-2027 and warned of 'agent washing'; a Swiss barometer (survey June–September 2026) found 36% of firms using agents and widespread measurement gaps; operational anecdotes include Klarna's February 2024 claim it replaced 700 agents and May 2025 rehiring, and Anthropic's Project Vend (27 June 2025) ran a month-long office shop that lost money. IA Académie recommends treating autonomy in three degrees (prepare with human validation; act alone on low-impact tasks; act externally only after measured error rates), logging every action, limiting access, and measuring baseline indicators before rollout.
HUMXN is offering free plumbing, electrical, HVAC, air-duct and carpet cleaning in exchange for allowing workers to wear capture devices inside homes; the program will run in Minneapolis, Chicago and Miami, the company says.
HUMXN says it will pay for qualifying home services and record technicians using wearable capture technology to collect real‑world training data for AI robots; the lab announced pilot activity in three cities and framed the program as a way to capture technicians’ on-the-job decisions and movements across diverse homes and climates.
The Chinese American Semiconductor Professional Association (CASPA) holds its 35th annual conference and CEO Summit at the Santa Clara Convention Center on 23–24 October, bringing 39 speakers including the chief executives of Intel, Synopsys, Cerebras and SambaNova to discuss AI compute costs, architectures and the industry’s move beyond GPUs.
The Chinese American Semiconductor Professional Association (CASPA) holds its 35th annual conference and CEO Summit at the Santa Clara Convention Center on 23–24 October, bringing 39 speakers including the chief executives of Intel, Synopsys, Cerebras and SambaNova to discuss AI compute costs, architectures and the industry’s move beyond GPUs. The programme includes a keynote from Credo CEO Bill Brennan titled "$100/Watt: The Evolving Economics of AI Infrastructure", panels on industry outlook and next-generation AI computing, and a closing conversation on wafer-scale AI with Cerebras CEO Andrew Feldman; tickets are sold through Eventbrite with prices listed.
An E&P Quick Read piece states that newsrooms have amassed decades of reporting and that artificial intelligence (AI) is creating new opportunities for publishers to unlock that archive, rediscover institutional knowledge and surface stories within past coverage.
An E&P Quick Read piece states that newsrooms have amassed decades of reporting and that artificial intelligence (AI) is creating new opportunities for publishers to unlock that archive, rediscover institutional knowledge and surface stories within past coverage.
A study of 32 news organizations across 10 South American countries (2023–2025) finds generative AI moved from experimentation after the public launch of ChatGPT in late 2022 into an operational element of journalism, concentrated in audience mediation, production support and political-institutional roles.
A study of 32 news organizations across 10 South American countries (2023–2025) finds generative AI moved from experimentation after the public launch of ChatGPT in late 2022 into an operational element of journalism, concentrated in audience mediation, production support and political-institutional roles.
Hevolus reports that PotentIA Edu, a project developed by Exlea and Hevolus, introduces artificial intelligence into the immersive platform Briedoo and will be presented at the stand.
Hevolus reports that PotentIA Edu, a project developed by Exlea and Hevolus, introduces artificial intelligence into the immersive platform Briedoo and will be presented at the stand.
Sanxingdui: Future Memories is a 90‑minute film produced with ByteDance's Jimeng AI by Bona Film Group's AIGMS centre that opens nationwide in China on 23 October; the National Film Administration granted a public screening licence on 28 April and the release date was announced on 27 September.
Sanxingdui: Future Memories is a 90‑minute film produced with ByteDance's Jimeng AI by Bona Film Group's AIGMS centre that opens nationwide in China on 23 October; the National Film Administration granted a public screening licence on 28 April and the release date was announced on 27 September. (claims: production method, length, opening date, licence date, announcement date) The film was approved under the regulator's animated feature category because no formal category exists for AI‑generated features, and the project involved 'more than a hundred people over roughly two years' with all characters described as original digital creations. (claims: regulatory filing as animation, lack of category, team size and duration, character creation) Observers and the publisher note this precedent may harden into practice and that classification affects crediting, review standards and audience expectations. (claims: precedent and industry implications, quoted WeChat commentary)
AMD announced on 28 September it will acquire Fei‑Fei Li’s World Labs in an all‑stock deal valued at about $8.2 billion, expected to close by the end of 2026 and subject to regulatory approval.
AMD announced on 28 September it will acquire Fei‑Fei Li’s World Labs in an all‑stock deal valued at about $8.2 billion, expected to close by the end of 2026 and subject to regulatory approval. World Labs’ Marble generates walkable 3D worlds but has limits compared with captured reality.
AutoVRse announced on 28 September that its VRseBuilder platform can reduce custom VR development time by up to 93% by converting procedures into editable VR modules and integrating AI assistants with Unity; the company cites internal project comparisons but has not published a breakdown of the underlying data or project count.
AutoVRse announced on 28 September that its VRseBuilder platform can reduce custom VR development time by up to 93% by converting procedures into editable VR modules and integrating AI assistants with Unity; the company cites internal project comparisons but has not published a breakdown of the underlying data or project count.
At Meta Connect on 23 September 2026, Meta unveiled roughly 100 g Meta VR Glasses (with processor, battery and storage in a separate puck), announced camera‑free Ray‑Ban Meta Audio glasses, and published a Private Processing cloud architecture for AI glasses; developer publishing remains restricted during the Developer Preview and UK trials of Ray‑Ban Display are available now, while VR Glasses are a 2027 purchasing prospect.
At Meta Connect on 23 September 2026, Meta unveiled roughly 100 g Meta VR Glasses (with processor, battery and storage in a separate puck), announced camera‑free Ray‑Ban Meta Audio glasses, and published a Private Processing cloud architecture for AI glasses; developer publishing remains restricted during the Developer Preview and UK trials of Ray‑Ban Display are available now, while VR Glasses are a 2027 purchasing prospect.
Abel Márquez, Lead Technology Product Manager at Wolters Kluwer Tax & Accounting Spain, describes a transition from pilots to deployment of AI embedded in business processes, highlights use cases in labour management and knowledge‑intensive tasks, and stresses human‑in‑the‑loop supervision and responsible‑AI principles.
In an interview, Abel Márquez says many Spanish organisations have moved beyond isolated tests and now seek concrete, productivity‑focused AI use cases; Wolters Kluwer presents an ‘expert’ AI integrated into workflows for labour management that works with real customer data and keeps final decisions with professionals.
Manuel Navarro reports that HP markets the EliteBook X G2i as a laptop family that integrates on-device and software AI features — from Poly Studio collaboration tools to Copilot in Windows — intended to cut interruptions, speed routine tasks and ease enterprise adoption of intelligent workflows.
The HP EliteBook X G2i is presented as a productivity-focused laptop line that bundles AI-driven meeting and collaboration features (noise cancellation, background blur, voice adjustment) integrated with Poly Studio technologies, Intel Core Ultra processors described as tuned for AI workloads, batteries of up to 6 cells, and Windows Copilot support — all framed as measures to reduce interruptions and simplify business adoption of AI.
Itera Process, a systems-integration firm with 25 years of experience, announced Frontier by Itera, a program that it says aims to accelerate adoption of AI-native and agentic architectures while providing governance, compliance and resilience through managed services and modular accelerators.
Frontier by Itera bundles maturity assessment, a Control Plane for traceability, an agent-focused ADLC, production-ready Frontier Swarm on AWS and a Quick Adoption methodology; Itera frames governance as a precondition for safely scaling agentic AI and offers 24×7 managed services to operationalize that governance.
A qualitative study published on revistavirtualis.mx reports that generative artificial intelligence (IAG) adds automated recombination of styles, hyperrealistic image generation and real‑time personalization to meme creation. The research documents an interview with Eduardo Salvatori of the Instituto Nacional de Bellos Memes and states that Salvatori combines AI outputs with prior human editing and later curation, shifting between levels of IAG autonomy.
The article—based on a qualitative, critical approach and an interview with Eduardo Salvatori—finds that IAG increases iteration and automates processes of resignification in memes, while enabling creators to externalize aesthetic processes; Salvatori reportedly operates at IAG autonomy levels 3, 4 and 5 and emphasizes human editing before and curatorial handling after AI use.
A German trade article reports that the Bavarian audio industry has moved beyond experiments and now uses AI in everyday editorial work, including a feasibility study with Antenne Bayern that produced a chatbot using live transcripts via a RAG architecture and multi-tiered safety filters.
XPLR Magazine reports four deployed AI examples from the Bavarian audio sector; among them, a KI.M–Antenne Bayern feasibility study created a chatbot that integrates live program transcripts with a Retrieval-Augmented Generation architecture to answer up-to-date listener queries, and the project included editorial filtering mechanisms intended to block polarizing or out-of-context requests.
Orange Business hosted (with Axys) a webinar, available as a replay, that presents using AI to bridge Finance and ESG data to improve reliability, speed and compliance, and offers practical guidance including use cases (deduplication, standardisation), automated data-quality approaches, a ROI roadmap, and a nine-pillar framework.
Orange Business hosted (with Axys) a webinar, available as a replay, that presents using AI to bridge Finance and ESG data to improve reliability, speed and compliance, and offers practical guidance including use cases (deduplication, standardisation), automated data-quality approaches, a ROI roadmap, and a nine-pillar framework.
Karsten Wellert describes a deliberate strategy of using AI for testing, virtual listener personas and chatbot support while keeping on-air voices and editorial decisions human and transparent.
In an interview, Antenne Bayern’s CDO Karsten Wellert says the station uses AI as a backstage tool: teams experiment and deploy only after benefits are clear, AI personas — trained on market-research data — give fast, virtual listener feedback, multiple models are combined by agents, and the live on-air program remains entirely human; chatbots identify themselves and offer opt‑outs.