Navigating GenAI Innovation in 2025 with Data Streaming
Welcome to the third issue of Follow the Stream! As transformative as the past year has been for Generative AI, 2025 promises even greater advancements across industries. This scroll explores predictions and emerging trends shaping the AI landscape in 2025, while also spotlighting notable success stories from 2024 that set the stage for what’s next.
The State of Generative AI in the Enterprise: Moving From Potential to Performance
The Deloitte AI Institute has been conducting a survey series tracking the pulse of GenAI adoption, scaling, and trends throughout 2024. The latest findings from its Q3 insights report focus on critical considerations such as data foundations, governance, risk, and compliance. One key finding highlights that “75% of organizations have increased their technology investments around data life cycle management due to Generative AI.”
Unlocking Data Value in the Age of AI and Data Streaming
AI's integration into the automotive sector is transforming customer experiences, as demonstrated by Seema Acharya from Mercedes-Benz at Current 2024. By leveraging data streaming platforms (DSP), Mercedes-Benz delivers hyper-personalized in-car experiences, enabling AI-driven recommendations such as optimal temperature, lighting, and route. Fueled by real-time data, these innovations drastically reduce time to market for new features, and ensure faster, data-driven decisions.
Tis’ the Season of Streaming, which means it's time to celebrate and innovate! Starting on December 9th, Confluent’s five-day learning event is an expert introduction to real-time and event-driven data flows. Day 2 is all about AI, with sessions on building dynamic GenAI apps and real-time knowledge bases for RAG applications.
Empowering the First Generation of Real-Time AI Startups
The AI Accelerator Program from Confluent for Startups is designed to support the next wave of innovation in artificial intelligence. This highly selective, 10-week virtual program seeks to collaborate with 10 to 15 early-stage AI startups that are building real-time applications utilizing the power of data streaming. As real-time AI continues to transform industries, Confluent is uniquely positioned to help startups harness the potential of data streaming to drive intelligent, automated decisions at scale.
Predictions 2025: An AI Reality Check Paves the Path for Long-Term Success
Forrester’s 2025 AI Predictions report highlights “the importance of marrying data and AI strategies,” as notable forecasts point to the pitfalls of prioritizing quick wins: many companies focused on immediate ROI will scale back their AI initiatives prematurely, risking long-term growth and missing out on future innovation.
Data Streaming in the Age of AI: Current 2024 Keynote
Jay Kreps and Confluent's top minds reveal the next game-changing evolution of data infrastructure, which effectively leverages AI and ubiquitous automation. Leaders in professional services, media, and automotive industries share success stories for how they harness the power of data products to revolutionize operations.
Gartner Predicts 30% of Generative AI Projects Will Be Abandoned After Proof of Concept by End of 2025
Many generative AI projects are at risk of being abandoned by 2025 due to issues like poor data quality, high costs, and unclear business value. Despite initial improvements in revenue and productivity, organizations struggle to justify the significant investments required for AI, especially since the financial benefits are often indirect and take time to materialize.
As a guest on Everyday AI podcast, Will LaForest, Global Field CTO at Confluent, discusses how data streaming is changing the way AI operates. Along with Jordan Wilson, he dives into how real-time data can enhance AI’s accuracy, enable faster decision-making, and provide stability for industries like finance.
GEP Boosts AI-Powered Supply Chain, Processing 1B Events per Month
GEP improved its AI capabilities by enabling real-time data integration across systems, which supported the development of AI-driven solutions like a GenAI chatbot for contextual insights. This shift also enhanced predictive AI, allowing GEP to predict outages 30 minutes in advance, leading to better decision-making and improved customer satisfaction.
AI Skilling: Intersection of Leadership Vision and Operational Strategy
As an emerging technology, AI presents enterprises with an upskilling challenge that requires input from both leadership and operational teams. Before rolling out initiatives, companies should ask “why” and “how” they are implementing AI, in order to clarify respective areas of responsibility, and to ensure that leadership and operational decisions are aligned.
Meta Releases AI Model That Can Check Other AI Models' Work
Meta has released new AI models, including a "Self-Taught Evaluator" that uses a "chain of thought" technique and AI-generated data for training. This model aims to improve accuracy in complex tasks and could pave the way for autonomous AI agents capable of self-improvement, cutting out the need for costly human feedback.
Billion Dollar Frauds: Unleashing the Power of Confluent’s Streaming AI to Combat Vishing
A recent scam where a Hong Kong branch manager lost $35 million to a deepfake voice emphasizes the urgent need for better protection against vishing. GoodLabs Studio is combating this by using Confluent and Apache Flink®'s Streaming AI to analyze real-time audio from Genesys, and match voices with established voiceprints. The strategy improves model accuracy through continuous training and enables rapid detection of vishing attempts in an effort to mitigate financial losses.
The 10 Biggest AI Trends of 2025 Everyone Must Be Ready for Today
In 2025, AI legislation will focus on combating discrimination and disinformation, while efforts to navigate a post-truth world will address the rise of AI-generated fake content. AI's role in cybersecurity and defense will expand, helping detect and counter evolving cyber threats. Additionally, sustainable AI is expected to drive both energy-efficient data centers and applications that reduce environmental impact across industries.
AI voice cloning is challenging the reliability of voice biometrics security widely used by financial institutions. To combat these threats, GoodLabs designed a vishing penetration test strategy that periodically assesses and enhances voice biometric systems. The solution uses advanced detection layers, including acoustic analysis, linguistic pattern recognition, and behavioral anomaly detection, to differentiate authentic voices from AI-generated ones.
Why Event Streaming Is the Backbone of Real-Time GenAI
Generative AI applications rely on timely and accurate data, which traditional data lakes and SaaS platforms struggle to provide. While data lakes simplify governance, they fall short in providing the immediate, well-contextualized data streams required for AI-driven applications like virtual assistants. Event-streaming platforms bridge this gap, enabling the continuous flow of up-to-date, contextual data from various sources to power responsive and reliable AI solutions.
The Digitalization of Airport and Airlines With IoT and Data Streaming
“The vision for a digitalized airport includes seamless passenger experiences, optimized operations, consistent integration with airlines and retail stores, and enhanced security.” Schiphol Group’s real-world implementations showcase how data streaming can optimize everything from predictive maintenance to personalized passenger services, ultimately driving automation and innovation in the aviation industry.
To address the challenges of data management in AI, Confluent and MongoDB emphasize the importance of making contextualized, trusted, and current data readily available to LLMs. The combination of Confluent and MongoDB enables the seamless flow and management of high-quality, real-time data, ensuring accelerated business transformation and responsive AI.
Penske Unlocks Vehicle Uptime & Supply Chain Visibility with Data Streaming and AI
Penske collects vast amounts of data from its over 400,000 vehicles, using sensors and IoT devices to monitor real-time metrics like location and engine status. AI-powered diagnostics analyze this data to predict vehicle failures before they happen, minimizing breakdowns and improving fleet management. The combination of real-time data streaming and AI allows Penske to improve supply chain visibility, and strengthens its ability to meet customer demands for on-time delivery.
Predictions 2025: GenAI Reality Bites Back For Software Developers
While AI assistants are rapidly gaining traction among developers, this capability is “only the beginning of the GenAI software development revolution.” Forrester’s predictions for 2025 suggest that 50% of enterprises will move away from individual best-of-breed tools in favor of comprehensive DevOps platforms, while at least one organization will attempt to replace 50% of its developers with AI—but fail.
How to Build GenAI Apps Faster with Confluent’s Connectors
While GenAI excels with real-time, contextual data, many enterprises face hurdles due to siloed systems and outdated batch-processing methods. A data streaming platform combined with a robust connector ecosystem streamlines the creation of GenAI applications, enabling real-time knowledge bases, AI-ready pipelines, and continuous updates to vector databases like MongoDB Atlas.
LLM Series—How to Expand (And Widen) LLM Brain Power
Large Language Models (LLMs) like GPT are powerful machine-learning tools trained on vast data to recognize language patterns. However, they are stateless and lack access to proprietary data, which limits their context. Developers play a crucial role in refining LLM outputs by ensuring the right data is available and accessible, thus enabling a more democratic, informed, and actionable approach to business decision-making.
Flink AI: Real-Time ML and GenAI Enrichment of Streaming Data with Flink SQL on Confluent Cloud
Flink AI integrates machine-learning models with real-time data streams, allowing businesses to enrich data dynamically. For instance, in a customer review analysis use case, Flink uses a remote AI model like GPT-4 to classify sentiment and generate ratings in real time. This solution enables businesses to respond to customer feedback with actionable insights, such as product availability or alternative suggestions.
Data-Driven HR: Exploring AI-Powered Chatbots and the Future of Employee Engagement
HR professionals are exploring AI tools with mixed reactions, balancing excitement and caution. Colleen McCreary, Confluent’s Chief People Officer, has created “AI Colleen,” a chatbot built from her past emails, blogs, and recordings, to test AI's potential in managing HR tasks. While McCreary acknowledges AI’s efficiency for handling routine questions, she warns against its use for complex, emotional, or privacy-sensitive issues, where human empathy and context are essential.
“According to IDC, French retailers see ROI from AI in just 14 months, generating an average profit of €3.45 for every €1 spent.” Although leaders like Carrefour and La Redoute have already started integrating GenAI solutions for tailored recommendations and intelligent chatbots, siloed and unreliable data in retail still hinders broader progress. To unlock AI’s full potential, retailers are shifting their focus to real-time, integrated data strategies, leading to accurate and accessible insights.
