What happens when an entire city becomes a neural network for ideas?
San Francisco’s Tech Week 2025 was more than a gathering; it was an experiment in collective intelligence. Thousands of decentralised events, hosted across the city by the a16z community and beyond, revealed not just what the technology industry is building, but why.
For me, and for DataSapien, this week reaffirmed something essential: The future of intelligence is personal, private, purposeful, and evolving toward the edge.
The Shift to Small Language Models and Edge AI
While large language models dominated headlines, the most transformative ideas were happening quietly, at the edge.
At an SF Tech Week event aptly titled “This Time It’s Different”, Professor Daniela Rus of MIT and CEO of liquid.AI unveiled a new class of Small Language Models (SLMs), designed by referencing the efficiency of organic synapses. Their resulting SLMs are ten times faster and more precise than Google’s latest SLMs. Not because they are larger, but because they are smarter by design. We’re excited to start deploying and benchmarking them with the DataSapien AI SDK and Orchestrator!
The next morning, at the personal.AI showcase, I saw more glimpses of what genuine “edge intelligence” will look like. Local private AI and agents for enterprise productivity, along with the NVIDIA-powered’ AI data centre in a box’. Both reflect the same philosophy that drives our work at DataSapien. We are fellow travellers, building in adjacent spaces but moving toward the same horizon: a world where intelligence happens privately, on-device, and in service of individual customers, patients, citizens – humans. The cloud is not disappearing, but the ecosystem is evolving, with Intelligence becoming smaller, closer, and far more personal.
The takeaway: As Large Language Models (LLMs) mature and commoditise, they will inevitably miniaturise, enhance efficiency and become personal. AI will follow the same path as compute; from Mainframes to Mobile phones, but in months rather than decades. AI will not live in the cloud, but with us and for us, and increasingly, on us.
At DataSapien, this is the world we have been building towards from the beginning – providing the developer platform where Zero-Shared Data fuels on-device personalisation, and where intelligence operates privately, for the benefit of the individual; customer/patient/citizen.
The AI Boom-Bust Cycle: What Comes Next?
Conversations about the AI boom-bust cycle were impossible to avoid this week, and rightly so. In one particularly candid exchange, Jerry Kaplan, long-time entrepreneur, Silicon Valley icon and AI historian, said it best:
“Yes, this time it is different… Just like the last time.” — Jerry Kaplan
That line drew laughter, but also a moment of reflection. For many decades, the cycles of exuberance, consolidation, and recalibration are as much a part of AI’s evolution as its breakthroughs.
Yet beneath the noise and excitement, the most interesting conversations looked beyond the current hype. They explored the data and AI world we will inhabit next – how decentralisation might coexist with regulation, how personal AI could redefine ownership of intelligence, and how systems might adapt to human variability rather than smooth it away.
These are not questions of scale, but of ethics, context, and consequence. The P-values of our shared AI experiment – those small but significant signals revealing where truth and trust hold, and where they fracture – are both concerning and awe-inspiring in equal measure. Which leads us to…
Trust: The Quiet Infrastructure of Intelligence
If one theme resonated for me more deeply than any other at SF Tech Week, it was trust.
At the Trust on Toast breakfast, Annabel Pemberton led a thoughtful discussion on how trust must be designed into intelligent systems from the start, rather than added later.
Later, over coffee with Richard Whitt of the GliaNet Alliance, we discussed the idea of AI fiduciaries. Companies sign the GliaNetAlliance fiduciary pledge to act with responsibility towards their users, a concept soon to be reinforced by the option of formal accreditation. DataSapien is proud to be the first Fiduciary pledge signatory. To discover more, check out the GliaNet website.
In a world already crowded with what some have described as #AISlop, Annabel and Richard’s work stood out for its clarity and practicality. Offering a path toward an AI-powered world that is accountable, explainable, and aligned with human intent.
At DataSapien, this is more than validation. It speaks to a shared vision. Our Personal AI SDK and Zero-Shared Data architecture are built on a single conviction: the most intelligent systems are those that respect and best understand their users.
This photograph on the highway out to the airport is, perhaps, a fitting metaphor for the week. In San Francisco, you could sense both the reasons to pause and the impossibility of doing so. The “Stop AI” reaction felt by some in society may well grow as the possible impacts of AI become more tangible. Yet everywhere the reality is clear: The arrival of AI into the everyday lives of billions of people is already well underway, and we are just at the beginning of significant societal shift. Conversations about trust, regulation, and human alignment continue, but the choice that remains is how we build, not whether we do. This is why trust, purpose, and responsible design matter more than ever. Also, the community of builders delivering it.
Finding the Tribe
Beyond the panels and prototypes, what truly defined SF Tech Week, for me, was community.
At the GBX Annual Gala, surrounded by British founders (and a few honorary Antipodeans 😉) shaping the Bay Area start-up scene, conversation turned from capital to conscience. The evening highlight was a fireside chat with Al Gore, reflecting on human purpose and planetary responsibility in the heart of Silicon Valley.
Throughout the week, often five or six events a day, one common current ran through every discussion: energy, intent, and purpose. Builders building. Thinkers testing. Dreamers doing.
I left with dozens of new contacts, friends, clients and collaborators (more on these soon) – our growing tribe. More importantly, a renewed sense of the better world that we are helping our clients to build, beyond the bubble.
Where This Leaves DataSapien
- Edge Intelligence Is Here. The rise of Small Language Models confirms that the next wave is happening on-device, not in the cloud.
- Trust Is the New Infrastructure. Fiduciary-grade transparency will fast become the standard for sustainable AI.
- Community Fuels Progress. Innovation is never a solo act. It is the outcome of shared conviction and collective energy.
SF Tech Week 2025 was a reminder of why we founded DataSapien: to enable organisations to give individuals control of their data, intelligence, and digital outcomes.
The world of AI is not just scaling up; it is scaling down, to the personal, exactly where DataSapien enables brands and publishers to deploy.
If you were part of SF Tech Week, I would love to hear what stayed with you; the models, the ideas, or perhaps, like me, the people who made it all real. Feel free to connect, comment or DM me on Linkedin.
#SFTechWeek #EdgeAI #PrivacyTech #TrustInAI #DataSapien #PersonalIntelligence