Brightsides.uk

Tag: artificial-intelligence

  • Confluent’s Predictions for GenAI in 2026

    According to a recent survey, 68% of IT leaders recently attributed data silos as a major impediment to AI success. To adapt, companies will need to take these steps: 

    • Exposing agent-safe APIs 
    • Adopting tokenised payment protocols 
    • Making real-time product data available, to forestall transaction bottlenecks down the pipeline for an “always-synchronised commerce layer.” 

    Confluent’s 2026 forecast explained, 

    “Companies will need to figure out how to optimize sales and marketing for the machines that will increasingly do the decision making, and in some cases, ultimately the buying. If you think human customers are fickle, machine customers can be ruthless: they have zero patience for latency, no brand loyalty, and can switch vendors mid-transaction whenever a better offer appears.” 

    Despite security concerns about the open-source protocol being shared with participants, who may not have been appropriately vetted, “the gravitational pull towards a single, easy open protocol that reduces friction and developer overhead will prove irresistible in 2026.” 

    Although it concedes that “other competing standards like Agent2Agent (A2A) and Agent Communication Protocol (ACP) continue to vie for relevance in agent-to-agent communication,” adopting the Model Context Protocol (MCP) framework will ensure flexibility in uptake of contributing LLMs, ensuring that your data vendor is fit for purpose and allowing trade-in for alt vendors. 

    The report asserts that context engineering will mainstream in 2025, whereas 2024 saw the roll-out of agentic AI. Context engineering is designed to iterate on data with priority schemas at the forefront of the analytic process, while not overburdening the LLM with run codes that slow down or impede the accuracy of its performance, in a continuous evaluation process guided by context. By using pre-fitted models, you are able “to refine logic, and adjust live through prompts, rules, and context.” 

    More offloading of queries to caches or additional databases will be mandated, to enable systems to cope with the data usage levels in 2026. “Now is the time to implement change data capture (CDC) pipelines and ensure data is flowing in near real time,” as agentic queries are run on a more comprehensive scale than those overseen by human actors and day-to-day data hygiene tasks are increasingly automated. 

    Confluent also stressed the vital role of cyber security in 2026, with many technical leaders citing it as a core priority. Current forecasts for global cyber crime account for losses of about £12trillion, although the uptake of AI by threat actors could bump the figure as high as $18trillion p.a. Attack volume may actually double, making scenario modelling and penetration testing essential to containing registered threats and prioritising responses according to business impact, to isolate contaminated divisions stacked by tiers according to priority. Cyber security managers need to be able to pivot around a threat, using segregated “high volume, low-latency data infrastructure for instant analysis and automated response.” 

    In the context of data governance, for safeguarding standards to be upheld, ensuring data trust, quality and lineage, data protection officers need to be proactive about monitoring reuse of sensitive data and linked queries run against data tables where access is permitted based not only on shared values, but also on overlapping fields in segregated data tables where bulk processing requirements can override usage rules. The report stated that 84% of technical leaders recently called data management and governance a top-tier technology priority. 

    Confluent is pushing its cold data siloing solution, Iceberg, as a way of meeting requirements for data reuse. 

    “ Iceberg is poised to lead this transformation, with continued maturity in its Puffin metadata format, advancements in data compaction and sorting, and emerging row-level lineage capabilities. Compared to other open table formats, like Databricks Delta Lake (which is optimized for “hot” read performance), Iceberg’s merge-on-read and partitioning design make it particularly well suited for cold data workloads.” 

    Durable execution engines will be the smart investment choice, with event-driven responses to retries, timeouts and compensations. LangGraph and Pydantic AI have pioneered the adoption of fault-tolerant frameworks, which allow for solution via live diagnostics. This enables complex patterns, such as those run by Sage and CQRS, to transition from microservice code into work-flow-as-code. 

    “At the same time, Kafka will remain the scalable event backbone and Flink a high-throughput low latency processing and real-time analytics engine.  

    The report concluded with a call to CIOs not to overwrite or override the existing infrastrucutre, but to work with GenAI to adapt the system to changing usage needs. 

    “Executives should begin to re-analyze the cost-risk for legacy system modernization with GenAI, leveraging specialized integrators for generative code translation and understanding. Developers should focus on migrating legacy messaging systems (like JMS) to modern event-driven architectures, using AI tools to accelerate the process and transform hard-to-maintain legacy code into new, valuable capabilities.” 

  • Microsoft’s annual report demonstrates continued AI innovations available across the income spectrum; and its commitment to diversity and inclusion, and cyber security

    Microsoft announced a record-beating amount in annual revenue of over $245 billion, a 16 percent year-on-year increase, with operating income up 24 percent at more than $109 billion. 

    As of June 30, 2024, $10.3 billion remained of the $60.0 billion share repurchase program which commenced in November 2021.  

    The last reported dividend was 14 December 2023, where Microsoft paid out $0.75 per share. Whilst the earnings per share compared to the S&P 500 and Nasdaq 500 shows it consistently beat both indexes, shareholders await the revelation of the dividend for Q1 2024. The total of the last dividend payout amounted to $5,574. 

    Fair Market Value (FMV) of actively traded shares amounted to $349.91 correct as of June 2024. A comparison of 5 year cumulative total return puts the calculation for the NASDAQ Computer Index at $331.2 and the S&P 500 Index’s aggregated return since 2019 at $201.5. These figures represent the net return on $100 invested on 6/30/19 in stock or index, factoring for reinvestment of dividends. 

    Its Diversity and Inclusion Report (hyper-link) highlighted its healthy workplace culture, whereby “Just as our culture has been critical in getting us to this point, it will be critical to our success going forward. At Microsoft, we think of our culture as being both input and output… For us, that means constantly exercising our growth mindset and confronting our fixed mindset—each one of us, every day. It is the only way we will succeed.” 

    Matched donations by 106,000 employees and employer amounted to a total $250million to almost 35,000 nonprofits across 111 countries, with the time spent volunteering by employees to charitable causes put at over 1million. Chairman and Chief Executive Officer Satya Nadella said in a CEO statement prepared October 2024 praised their non-profit oriented stakeholder engagement: 

    “I am deeply grateful for my colleagues’ dedication to making a difference. Together, we can continue to empower everyone around the world.” 

    In the context of AI developments, Microsoft was pleased to announce the roll-out of CoPilot as an add-on in both Business and Home and Personal versions of Microsoft Office 365…. 

    Co-Pilot for professionals is underpinned by secure GitHub repositories. One case study of Brazil’s largest bank, Itau, has since the application was rolled out across terminals seen a 68% increase in deployment speed and a 75% rate of code re-use demonstrating continued internal use-cases. The organisation bore witness to a 93% increase in deployment speed since it linked to the new GitHub repositories. This, says the case study write-up, helps them allocate more time to developing new systems, with server connectivity assured. 

    In Kenya, where much of the population does not have easy access to a bank account and no way to demonstrate their credit score, street vendors used the deployment M-Kopa, a social enterprise using Azure ML to do its forecasting utilising large language models for leads gen of financially inclusive loans issuance. 

    Microsoft’s annual report said, 

     “We offer leading frontier models, thanks to our strategic partnership with OpenAI. With Phi-3, which we announced in April, we offer a family of powerful, small language models. And, with Models as a service, we provide API access to third-party models, including the latest from Cohere, Meta, and Mistral. In total, we have over 60,000 Azure AI customers, up nearly 60 percent year-over-year… 

    This year, we also introduced Copilot Workspace, a Copilot-native developer environment, which helps any developer go from idea, to code, to software—all in natural language.” 

    Its offering Power Platform provides LLM accessible to all users, whether their use case is developing a website, automating workflows, or building a website. Year-on-year there was a net 40% increase in the user base of Power Platform, to a monthly figure of 48million users. 

    Data processing is dependent on large secured data lakes and effective connectivity when undergoing data warehousing. Microsoft said its Microsoft Intelligent Data Platform enabled business intelligence spanning storage siloes with vector embedding driving access to AI capabilities. Its new AI-powered, next-generation data platform Microsoft Fabric has a paid user base of 14,000 customers who can leverage and action their data insights within a unified SaaS fix. 

    It said that even its Microsoft Teams platform was seeing a huge up-ticc in popularity, enabling encrypted communications for a secure workplace environment, – Teams Premium surpassed 3 million seats, up nearly 400 percent year-over-year.  

    Professionalizing its GitHub Co-Pilot offering, which is used by 60% of Fortune 500 companies to streamline and increase velocity on workflow desks, has resulted in for example the Dynamics 365 Contact Center being able to integrate existing legacy infrastructure of CRM systems with advanced AI capability. 

    New use cases of targeted business applications have been found in the healthcare arena – with the DAX Co-Pilot, more than 400 healthcare organizations are increasing physician productivity and reducing burnout. On average, clinicians save more than five minutes per patient encounter. And 77 percent say it also improves documentation quality. 

    Its commitment to cyber security is evidenced by collaboration across systemically important IT service providers. “ At the Munich Security Conference in February, we came together with others across the tech sector and pledged to help prevent deceptive AI content from interfering with global elections. As part of this pledge, we have worked to empower campaigns, candidates, election officials, and voters to understand the risks of deceptive AI in elections and to take steps to protect themselves and democracies. To date, we’ve conducted deepfake trainings in over 20 countries. And our corresponding public awareness campaign has reached over 355 million people.”