Presenters

Source

London’s AI Moment: Powering the Next Wave of Innovation with MongoDB 🚀

London is buzzing with AI innovation, and MongoDB is at the heart of it all! From groundbreaking AI labs to established financial giants, companies in the UK are leveraging MongoDB’s unified platform to build the future. This isn’t just about faster processing; it’s about smarter processing, more accurate retrieval, and ultimately, trusted AI agents that can transform businesses.

This post dives into why London is a special place for this AI revolution, the challenges and opportunities in building AI agents, and how MongoDB is empowering developers and enterprises to not just participate, but lead this transformation.

London: A Hub for AI Revolution 🇬🇧

London isn’t just a city; it’s a historical powerhouse of innovation, from the Industrial Revolution to the current AI era. CJ Desai, our speaker, emphasizes London’s unique position in the AI landscape.

  • Frontier Labs: A leading frontier lab company, with its founding engineer based in London, chose MongoDB to build its AI solutions.
  • Enterprise Bet: A UK-based financial services company made a significant early bet on MongoDB in 2017-2018, building mission-critical applications on the platform.
  • AI Native Success: 11 Labs, a phenomenal AI-native company headquartered in London, is also powered by MongoDB, showcasing the city’s vibrant AI startup scene.

CJ likens the current AI revolution to the industrial revolution, but predicts it will be much bigger and move at a much faster pace.

Navigating the AI Agentic Architecture 🤖

The journey to building intelligent AI agents involves understanding their core components and challenges.

  • The Three Pillars: Agentic architecture rests on three key legs:
    1. Data Layer: The foundation for agent intelligence.
    2. LLMs (Large Language Models): The brains behind the operation.
    3. Harness & Orchestration: The glue that connects everything.
  • The Data Layer is Key: The argument is strong: start with the data layer. Ensure your data is ready for agents to run at scale and in real-time.
  • OLTP vs. OLAP: For real-time business decisions and true agent intelligence, OLTP (Online Transaction Processing) is the “high ground.” OLAP (Online Analytical Processing) databases, while useful for looking back, often have delays.
  • MongoDB’s Unified Platform: MongoDB offers a single, unified data layer integrating transactional data with text search, vector search, and embeddings, providing all the functionality needed to build scalable, real-time agents.

Core Strengths for Agentic AI: Accuracy, Run Anywhere, Performance 🎯

MongoDB’s strengths are crucial for building successful AI agents:

1. Accuracy: Beyond the LLM 💡

  • The Retrieval Issue: Many believe AI inaccuracies stem from LLMs. However, the real culprit is often a retrieval issue – how data is searched and fetched from various sources.
  • Financial Times Example: The Financial Times leverages MongoDB to unify its vast data archives, enabling users (and agents) to search historical articles with natural language queries, ensuring relevant and accurate results. This is vital for trust, especially in data-sensitive industries like finance, healthcare, and public sector.
  • Unified Functionality: MongoDB’s single platform provides text search, semantic search, and embeddings to solve this retrieval problem, ensuring agents deliver clean, relevant results.

2. Run Anywhere: Data Sovereignty & Resiliency 🌐

  • Operational Resiliency: In today’s world, regulators and businesses demand high availability and resiliency.
  • Data Sovereignty: Global firms must adhere to data sovereignty laws, making it critical to control where data resides.
  • Multi-Cloud Flexibility: MongoDB allows clusters to be distributed across multiple clouds (AWS, GCP, Azure) and even on-premises. This is crucial for:
    • Capacity: Avoiding limitations with specific cloud providers.
    • Resiliency: Mitigating risks from hyperscaler outages.
    • Cost Optimization: Leveraging specialized AI services from different providers.
  • Model & Cloud Agnostic: MongoDB is designed to work with any LLM and run on any cloud, offering unparalleled flexibility.

3. Performance, Scale, and Availability: The Unsleeping Foundation 🔋

  • Agentic Scale: As agents interact with databases more than humans, the database must scale significantly for reads, writes, and transactions.
  • Foundational Priority: MongoDB prioritizes security, performance, scale, and availability to support both human and machine workloads.
  • Top UK Telco Example: A major UK telco uses MongoDB to create a persistent and performant memory layer for its 40 million customers, streamlining interactions across human and voice agents, ensuring customers don’t have to repeat themselves.
  • Continuous Innovation: MongoDB releases, like version 8.0 and 9.0, focus on accelerating innovation and delivering faster performance.
  • New Releases:
    • Atlas Auto Embeddings (Public Preview): Automatically converts raw text into embeddings for semantic search, simplifying data pipelines.
    • MongoDB 8.3 Release: Boasts significantly faster performance for reads, writes, and ACID transactions.

The Power of Embeddings and Reranking: Making Agents Trustworthy 🧠

Frank and Miko from MongoDB highlighted a critical challenge: trust. Many agents fail to reach production because they lack memory and accuracy, leading to inconsistent and wrong information.

  • Memory is Key: Agents need memory to perceive, plan, and act coherently across sessions.
  • Beyond LLMs: Bad AI is often a memory and retrieval problem, not solely an LLM issue. The LLM only acts on the information it’s given.
  • Embeddings: Convert data (text, images, audio) into numerical vectors that capture meaning. Proximity in this vector space indicates similarity, allowing agents to find relevant information beyond keywords.
  • Rerankers: Refine the results from embeddings, reordering them by relevance to improve accuracy. Think of embeddings as a wide net, and rerankers as the hand selecting the best catch.
  • Voyage by MongoDB: Offers industry-leading embedding and reranking models, integrated directly into MongoDB.
    • Performance: Voyage 4 models top Hugging Face’s retrieval embedding benchmark, outperforming Google and OpenAI.
    • Cost-Effective: Shared embedding spaces reduce cost and latency.
    • Seamless Integration: Natively integrated into MongoDB, eliminating the need to stitch together multiple systems.
  • Atlas Auto Embeddings: Automates the embedding pipeline, saving development time and operational effort.
  • Demo: The MongoDB Open: A compelling demonstration showed agents managing a tennis tournament, using embeddings and rerankers to personalize offers, redirect visitors during rain delays, and ultimately increase revenue and improve reputation by keeping visitors engaged. This showcased accurate retrieval and decisive action in production.

Investing in the Future: Founders, Europe, and AI 🇪🇺

Luciana Lixandru from Sequoia Capital shared insights into the investment landscape, particularly for AI companies in Europe.

  • Focus on Founders: Sequoia invests in great people who move really fast, possess an unfair talent advantage, and can become the best in the world at one specific, impactful thing.
  • European AI Ecosystem: London is a growing hub for AI innovation, attracting founders from across Europe and North America.
  • The “Mode” in AI: While software used to be a moat, in the AI era, deep vertical expertise and specialized talent are becoming key differentiators. 11 Labs, for instance, is recognized for its exceptional voice models.
  • Physical AI: The next frontier may lie in “physical AI,” encompassing robotics and defense, where specialized founders are building significant companies.

11 Labs: Revolutionizing Voice AI with MongoDB 🗣️

Alex Holt from 11 Labs discussed the company’s rapid growth and its strategic use of MongoDB.

  • From Lab to Product: 11 Labs evolved from a research lab building revolutionary text-to-speech APIs to a product company enabling businesses to create human-sounding voice agents.
  • Mission: To change how humans interact with businesses through technology, eliminating forms and hold times.
  • MongoDB for Agents: 11 Labs uses MongoDB not just for their core speech-to-speech models but also as the foundational data layer for their agentic architecture. This allows them to focus on model innovation without worrying about data management complexities.
  • Scalability: With 40 million agents deployed, MongoDB provides the robust and scalable infrastructure needed.
  • Operational Use Cases: Beyond customer-facing applications, 11 Labs leverages voice AI for operational tasks, such as autonomously verifying restaurant opening hours for delivery services.

The Decade of AI: A Transformation at Lloyd’s Banking Group 🏦

Ulku Rowe, CIO of Commercial Banking and Capital Markets at Lloyd’s Banking Group, shared her perspective on the AI transformation within a historic financial institution.

  • AI as a Decade-Long Journey: Lloyd’s views AI not as a year-long initiative but as a decade-long transformation, embedding AI across all aspects of the business.
  • Market Intelligence: LBG leverages its vast transactional data to provide aggregated, anonymous insights to businesses, helping them benchmark performance, identify new opportunities, and understand market dynamics.
  • People Transformation: Significant investment is being made in upskilling engineers and knowledge workers through programs like the “AI Academy” and executive programs with Cambridge University, ensuring AI literacy across the organization.
  • Ecosystem Partnerships: Lloyd’s partners with key players in the AI ecosystem, focusing on collaborative relationships that drive shared goals and innovative client experiences.
  • Holistic Approach: AI transformation is intertwined with infrastructure modernization, data management, and talent development. It’s about building a cohesive ecosystem where AI is the “icing on the cake” that pulls everything together.

The Future is Agentic, and MongoDB is Your Foundation 🛠️

The message is clear: the future of technology is agentic, and it’s moving at an unprecedented pace. MongoDB is committed to providing the foundational platform that empowers you to build, deploy, and scale these intelligent agents with confidence.

Whether you are a startup or an enterprise, MongoDB offers the accuracy, flexibility, performance, and scalability needed to navigate the AI revolution and build the next generation of transformative applications.

Appendix