PUBLISHED DATE: 2026-08-08 04:10:15
VIDEO TRANSCRIPT:
SPEAKER A:
So now you've heard how agentic AI changes journeys, processes and decision-making. What I want to talk about is what we learned along the way doing this for real, implementing agentic AI in enterprises, the scars. Not the edge cases, not the theory, just the patterns that show up every time you try to move from pilot to enterprise reality. Battle cards exist because repeating the same lesson over and over again is the most expensive way for any enterprise to learn. Why do scars repeat? What surprised us the most wasn't that things went wrong, it's how predictably they went wrong. Different industries, different teams had very similar patterns and the root cause was experience trapped in people's heads instead of the systems.
Let's dive into the scars one by one. Data is good enough until AI sees it. Data is good enough until AI starts reasoning across it. And that's where gaps start appearing, gaps in meaning and context surface quickly. You know, you see where context is badly defined or where the meaning is misinterpreted. And that's why context really matters, not just what the data says, but what it means and where it's come from. come from and when it should not be trusted. We now capture those conditions as battle cards so AI knows when to proceed, pause or to escalate where needed and bring a human in the loop.
Integration is always harder than planned. Integration starts simple but becomes fragile over time. Interoperability is really what matters the most and what we learned is that integration isn't really about systems, it's about interoperability over time. So with RPS agentic AI platforms, the orchestration and how our systems become interoperable is really at the core of how we design. our agentic AI architecture and ecosystem and that's paramount.
Adoption breaks before technology does so some of the toughest scars we've seen had nothing to do with how the AI is performing. Teams don't trust the output in some cases or leaders are not really aligned on what AI was allowed to do and expectations become unclear. And we now capture those moments, what created trust and what eroded it so future teams don't have to repeat the same mistakes over and over again and adoption becomes easier.
Pilots prove value but stall because of challenges in scaling to our large highly regulated enterprise like in financial services. And this is why adopting agentic platforms created specifically for enterprise use alongside experts that really understand your business is the most crucial to create an agentic workflow at scale. So that's something that we always make sure is in place. When we are designing for large enterprise-wide ecosystems, let's dive into some questions.
The answer is two mechanisms. First, battle cards are never there to execute, so they only inform the decision. AI is the system suggesting, however the humans are the ones deciding and that's the importance of the human in the loop. The second is that outcomes are continuously attached, so if a decision leads to a poor outcome, that signal becomes part of future recommendations and part of our feedback loop.
The effort is minimal if it's done correctly. We don't have to interview everyone, create massive work packages to create these battle cards. They're capturing decisions as they naturally occur. So they're embedded as part of our process and ways of working in reviews and escalations and exceptions. The discipline then becomes curation and less about extensive documentation.
The answer is that it strengthens both. Instead of policies alone, leaders can now look at which decisions were made, under what constraints and what outcomes, and really see the traceability in all of this. So this improves auditability and learning at the same time, supporting governance and compliance and auditability requirements in regulated environments.
To wrap this up. You can turn experience into reusable intelligence across your organization and integrate it into your AI factory and this leads on to our next episode where you'll see how the AI factory takes all of this, operationalizes it so memory learning and governance doesn't depend on people or heroics but on systems.