Winter Kickoff
November 2025 Instalment of Building AI for Impact
Hello, hello! We hope you’re having a wonderful start to the chilly winter season!
In this edition of Building AI for Impact, we’re taking a moment to reflect on the things we’re grateful for and diving deep into one of the most exciting frontiers in AI: agents. We’ve also been working on some exciting ideas around autonomous systems and their potential for optimising supply chains. Let’s jump right in!
What’s Inside This Months Deep Dive
Gratitude and Reflection: As we head into the holiday season, the Ergodic team is taking a moment to express our sincere appreciation for the community, collaborators, and supporters who have helped make this year truly remarkable. With Thanksgiving, in the US in mind, to our wider global outlook, we’re reflecting on the work we’re building together.
Agents: This month, we’re exploring autonomous agents…systems capable of perceiving their environment, making decisions, and taking action to achieve specific goals. How do agents actually work, and what role will they play in reshaping how enterprises operate?
Looking Ahead to December: For our local London followers, we have two events this week!! Take a look below and we would love to see you there!
As always, we are rounding out the newsletter with our reading list!
October kept us busy, and November has been no different! We’ve been hard at work, developing and preparing new insights for the bustling season ahead. Let’s dive into this month’s edition of Building AI for Impact!
Gratitude & Building in Community
As we approach the festive season and near the end of what has been an outstanding year, our team wanted to pause and offer genuine appreciation for the people and moments that have shaped our journey.
Thanksgiving is more than just about a roast dinner and pumpkin pie; it’s a time to recognise the people and relationships that make everything meaningful. We’re deeply grateful for the collaborators who have challenged our thinking, the companies that have trusted us with their complex systems, and the researchers who are constantly pushing the boundaries of what artificial intelligence can achieve.
So, here’s a heartfelt thank you to everyone who’s been a part of our journey. Whether you’ve attended our events, engaged with our platform, or simply believed in what we’re building at Ergodic, you’ve contributed to something truly impactful.
As we head into the final stretch of 2025, we’re energised by your support and excited to continue developing systems that drive real, measurable change.
Agents: What They Are and Why They Matter
For months, we’ve been emphasising the importance of comprehension and education within AI systems. Now, we want to take a closer look at a paradigm that could fundamentally reshape how that comprehension is deployed: autonomous agents.
So, what exactly is an agent? At its core, an agent is an AI system that can perceive its environment, reason about what’s happening, make decisions, and take actions to achieve specific goals. Unlike traditional LLMs that respond to queries or make static predictions, agents are dynamic. They operate continuously, adapt to changing conditions, and can break down complex problems into manageable tasks.
Let’s explore the difference with an example:
An LLM might tell you, “Demand is expected to increase by 20% next week,” based on your uploaded datasets.
An agent, on the other hand, could autonomously monitor system signals, detect the demand spike, initiate contingency plans, coordinate those plans, adjust inventory positioning, and even flag any edge cases requiring human judgment. The agent doesn’t just tell you what’s happening; it takes action.
But we believe the real opportunity lies in enterprise AI that doesn’t need to be fully autonomous to be powerful. The most compelling agents are those designed to work in partnership with human expertise. They handle the continuous monitoring, pattern recognition, and coordination across complex networks, while humans provide the judgement, strategic direction, and capacity to navigate novel situations that fall outside expected parameters.
Here are three key principles for agents in enterprise environments:
Transparency First: Agents should explain their reasoning and actions in human-understandable terms. For example, “I’m rerouting shipment X because Supplier Y has just lost two production lines,” is far more valuable than a black-box action log.
Human-Agent Collaboration: The most effective agents augment human capabilities rather than replace them. Agents handle scale and speed; humans handle nuance and judgement.
Safety by Design: Agents need clear boundaries, robust error handling, and the ability to escalate decisions when they encounter uncertainty beyond their training.
The agents that will thrive in enterprise settings aren’t the ones claiming full autonomy; they are those that are transparent about their limitations, clear in their reasoning, and genuinely designed to make humans more capable.
But, even beyond agents, and agent-human augmentation, there’s a piece of the puzzle still missing. Enterprises still face a fundamental unresolved problem: bringing the right people, the right perspectives, and the right agents together to converge on a shared truth and act quickly.
Automation alone doesn’t solve this. Most organisations have automated the identification of issues, but not the alignment required to resolve them: surfacing the right problems to the right stakeholders at the right time, combining human judgement with AI insights, driving shared understanding, and orchestrating cross-functional action. The solution is the next evolution of enterprise intelligence: multi-user, AI-augmented collaboration environments where multiple humans work within the same workspace, multiple specialised agents contribute monitoring, problem-solving, simulation, and planning, and a coordination layer ensures shared context, traceability, and alignment. All users will see the same real-time system state, problem-solving agents propose and rank hypotheses, and collaboration agents escalate open questions, and every insight is transparent and traceable.
Looking Ahead to December!
Today, December 2nd, Zubair Magrey will be at the Deep Tech Alliance Explore 2025 event here in London! We’re thrilled to have been selected to join the Deep Tech Alliance and to spend the day with the entire community connecting, sharing, and sparking new collaborations!
Then, later this week on the 4th of December, we’ll be showcasing at Venture Café London’s Thursday Gathering, in partnership with Thames Freeport: From Ideas to Impact: Collaborating for an Inclusive AI Future. The programme will explore the opportunities and partnerships necessary to make AI adoption inclusive, practical, and transformative across industries. Make sure to register and come along—we’d love to see you there!
Reading List
As always, here’s our reading list of the month!
More of Silicon Valley is building on free Chinese AI
University of Surrey researchers mimic brain wiring to improve AI - BBC News
Universities Must Reclaim AI Research for the Public Good | Stanford HAI
UK Budget 2025: Government Bets on AI and Startups - TechRepublic
[2510.25137] The Iceberg Index: Measuring Workforce Exposure Across the AI Economy
How Agentic AI is Transforming Enterprise Platforms | BCG
https://medium.com/@22amara.kenza/when-we-need-explainability-and-when-we-dont-326dc19d8577
November reminds us that the most innovation is built in service of something larger than itself. Whether it’s gratitude for community or excitement about technology that can truly make an impact, we’re reminded that artificial intelligence growth depends on who builds it, how they build it, and what values guide them.
That’s a wrap for this month! Thanks for being part of our community as we close out 2025!
Until Next Month,
The Ergodic Team