How We Agent

How We Agent

By Lyzr AITechnology
Download on the App Store

All How We Agent episodes

  • How Living Beyond Breast Cancer Is Using AI to Explain Lab Results to Patients

    Jean Sachs runs an organisation that reaches two million people a year with breast cancer information, supporting people trying to understand a complex diagnosis. Closing that information gap is exactly what AI should be good at and exactly where a confidently wrong answer does the most damage.

    This episode covers how she moves forward regardless: what they started with, what they won't touch yet, and why the first build was internal.

    Worth watching if you've had to introduce AI into a room where the sceptics have a point.

    16 min
  • How Awestruck Used an AI Audit to Find and Cut 360 Wasted Hours a Month

    Ryan Sprance co-founded Awestruck, which scaled from $106K in 2019 to $31M in 2025 without paid marketing. Along the way they found 360 hours a month going into manually formatting client reports and twelve tools worth building.

    A budget pacing tool that outperformed the commercial product it replaced. A reputation platform for 150+ clients, grounded in each brand's own voice and local facts. A commission app that removed 20 hours of monthly Excel work.

    30 min
  • How Growth Directors Gets AI Pilots Out of Pilot and into Production

    Ronan Langford's argument is that many enterprise AI programmes are stuck in what he calls the PoC Paradox: a steady stream of pilots that make existing processes marginally faster, while nobody at the top can describe what the company looks like on the other side.

    The argument here is to shift the emphasis: real-time benchmarking, simulations where autonomous avatars of regulators and unions stress-test decisions against supply chain and geopolitical shocks, and workflow redesign on sovereign-ready infrastructure.

    Worth a listen if your AI strategy is currently a spreadsheet of pilots nobody can connect to a future state.

    24 min
  • How Yugabyte Is Solving Agent Memory at the Database Layer

    Most teams building agents run into the same problem: each agent accumulates its own private memory, and the organisation quietly acquires a dozen new information silos.

    This episode is about addressing that at the database layer rather than the application layer. Yugabyte's new product, Meiko, gives agents shared memory and knowledge, so context compounds across agents.

    The more interesting part is what they've built on it internally worth a listen if your agents keep re-learning what the last one already knew.

    34 min
  • How Prophet Used Agents to Cut an 18-Month Delivery Cycle Down to Weeks

    Chan Suh leads digital at Prophet, whose client list runs Google, Netflix, Amazon and Marriott. The firm's core hasn't changed, but GenAI has collapsed a delivery cycle of six months research, six of strategy and six to launch into something closer to weeks.

    Their answer is Maia, a multi-agent OS built with Lyzr and structured like an internal app store, with agents carrying their proprietary consumer survey and 30 years of activation data.

    The shift in what clients ask for is the real headline: not a point solution, but a workbench they can run a process on.

    30 min
  • How Movate Used Agents to Cut Test Prep Time by up to 40%

    Devanathan Desikan runs Movate's momentum and innovation division, and made a call most teams get wrong: rather than spend two years building a platform, orchestrate on one.

    They built on Lyzr and shipped digital twins of roles - a coding twin, a QA twin - delivered as plug-ins inside Jira, VS Code, Slack, Teams and ServiceNow, so nobody has to context-switch to use them.

    The part worth borrowing is downstream: human-agent hybrid delivery let them move to outcome-based pricing, which meant retraining legal and pricing teams to underwrite what an agent commits to in an SOW.

    37 min