Last week, Superagentic AI proudly exhibited at ODSC AI West 2025 in San Francisco, the global innovation hub of AI. We showcased our pioneering work on Full-Stack Agent Optimization, connecting with attendees, partners, and speakers throughout the event. It was an incredible experience for our young company and a milestone moment for Superagentic AI.
Over three dynamic days, we connected with researchers, engineers, and practitioners from across the AI ecosystem in the Bay Area and around the world, sharing our vision of how the next generation of intelligent systems will be defined by continuous optimization across the entire Agentic AI stack. With the help of some amazing friends from the Bay Area, we set up an incredible booth that made the conference experience smoother. Superagentic AI wrapped up an incredible three days at ODSC West 2025 in San Francisco, where we exhibited for the first time and introduced SuperOptiX, our full-stack Agent Optimization framework. From hundreds of conversations to a packed startup pitch session, ODSC marked a milestone in our journey to shape the Agentic AI ecosystem.
Agentic AI Was Everywhere at ODSC AI
If there was one theme that echoed throughout ODSC AI and Silicon Valley, it was clear: AI systems are becoming Agentic. The community is now looking beyond individual models to AI agents, dynamic systems capable of reasoning, retrieving, acting, and improving through feedback. But as agents become more complex, the challenge shifts from creating them to optimizing how they work.
At Superagentic AI, we believe the next era of AI innovation lies in Full-Stack Agent Optimization, optimizing every layer of the agent lifecycle, from prompts and retrieval (RAG) to MCP tools, memory, and the entire context orchestration process.
That philosophy underpins our flagship Python framework, SuperOptiX, which helps developers and researchers build evaluation-first, optimization-core, and orchestration-driven agentic systems designed for reliability and adaptability.
Our Presence at ODSC San Francisco 2025
As part of the ODSC Exhibitor and Partner program, our team and friends helped set up the Superagentic AI booth, full of energy around Agentic AI infrastructure, where we introduced SuperOptiX to hundreds of conference attendees across three energetic days. The Superagentic AI booth was focused on knowledge rather than swag, with a striking black and white contrast in its design.
The audience response was phenomenal. Our booth saw a steady flow of data scientists, researchers, and AI/ML engineers interested in optimizing their AI agents for performance, adaptability, and scalability. Many were curious about how SuperOptiX goes beyond prompt tuning to deliver optimization across retrieval pipelines, tool interactions, and contextual memory systems.
We had engaging conversations with teams from Google, OpenAI, LangChain, Arize, Baseten, LlamaIndex, and many others actively exploring the Agent Optimization space. These discussions reaffirmed a shared belief that context engineering is just the beginning, and the real breakthrough lies in systematic, data-driven optimization of every agent layer.
Our booth also featured open-source tools such as DSPy and GEPA, which drew curiosity from attendees eager to learn how these tools play a role in Agentic AI Optimization.
Notable people we met at the ODSC AI Conference:
- Michelle Bonat from AI Squared
- Ali Arsanjani from Google Cloud
- Harrison Chase from LangChain
- Eno from Factory AI
- Shikhar Kwatra from OpenAI
- David Loker from CodeRabbit
Some were interested in our approach to Agent Optimization, while others showed strong interest in collaborating and building communities in the UK. We’ll be reaching out to thank them personally for making ODSC AI special. You can find all our booth photos and presence highlights here.
AI Expo and the Startup Talk by Superagentic AI
At ODSC AI, Superagentic AI delivered a startup pitch session titled “The Optimization Layer of Agentic AI: Beyond Context Engineering” on October 30th, 2025. In this talk, we explored how the next generation of AI agents will rely not only on better models but on continuous, systemic optimization where prompts, retrieval systems, memory, and context evolve through feedback and evaluation.
We introduced our new business model, Forward Deployed Agents (FDA), which enable companies to deploy agents on-premise and continuously optimize them using their own data. Although our talk coincided with tea time and book signings, we were thrilled to have a focused and curious audience. The talk was not recorded, but you can view highlights here.
The AI Expo was a major highlight for Superagentic AI. Our booth, featuring two large tables, a TV, and a digital display, stood out with its sleek black-and-white setup, a blend of digital and physical presence that drew consistent attention from attendees.
Booth, Conversations, and Collaboration
Running a booth at a major AI conference like ODSC AI was exhilarating. For Superagentic AI, it was the first experience exhibiting as a startup. As a solo founder, I was fortunate to have incredible friends from the Bay Area helping at the booth. Together, we engaged attendees, ran demos, and shared how SuperOptiX brings an optimization-first philosophy to Agentic AI.
The Superagentic AI booth became a hub of curiosity and collaboration, with constant discussions around evaluating and improving agent behavior in real-world systems. Many data scientists resonated with our optimization concepts, drawing parallels to their work with frameworks like PyTorch. While we couldn’t attend as many talks as we’d hoped, we’ve been actively following up with speakers, researchers, and attendees post-conference, many of these conversations are already evolving into collaborations.
Key Takeaways from ODSC 2025
- Agentic AI is everywhere: Almost every talk and conversation mentioned AI agents, their adoption, challenges, and practical use cases.
- Context Engineering is the new norm: Talks by LangChain, CodeRabbit, and others highlighted the move beyond prompt engineering toward Agentic Context Engineering.
- Value optimization of AI agents needs more understanding: The concept of full-stack Agent Optimization is still emerging, and we’re committed to leading this conversation as the field matures.
- Huge demand for AI training: There’s an enormous appetite for structured learning around AI agents, evals, and optimization. Superagentic AI plans to help meet this demand through our upcoming SuperRadar initiative.
Looking Ahead: The Journey Continues
ODSC AI San Francisco 2025 was more than a conference, it was a milestone in how the Bay Area AI community thinks about intelligence, systems, and optimization. For Superagentic AI, it validated our mission to make Agent Optimization and Agent Engineering core engineering disciplines for building the next generation of adaptive, self-improving systems.
We’re deeply grateful to the ODSC AI West organizers and everyone who visited our booth, attended our talk, or shared their ideas. Your enthusiasm and insights reinforced that we’re building in a space that truly matters. Special thanks to ODSC AI founder Sheamus and our amazing event guide Christina for their support.
What’s Next for Superagentic AI
- Expand SuperOptiX with richer evaluation and optimization features and collaborate with researchers and partners to push the boundaries of full-stack Agentic Optimization.
- Develop integrations and developer tools to make optimization accessible to everyone building in AI.
- Launch SuperRadar: our training and evaluation platform for Agentic AI and optimization.
- Exhibit and partner at more AI conferences across the Bay Area and globally.
If you’re exploring or researching in the Agent Optimization space, we’d love to connect and collaborate. Stay connected, explore Superagentic AI products and solutions, and reach out if we can help you adopt Agentic AI in your business. If you’re building in the Agentic AI space, let’s collaborate, explore SuperOptiX, or reach out to see how we can optimize your next generation of AI systems together with our new business model Forward Deployed Agents.