Episode summary
How Agents Can Work Together — With Vijoy Pandey
Big Technology Podcast
· · 37 min · Hosts: Ranjan Roy, Alex Stamos · Guests: Vijoy Pandey
Summary
Vijoy Pandey argues that enterprise AI is moving from isolated agents to connected multi-agent systems, but the hard part is no longer wiring them together. Cisco’s work through open-source efforts like Agency, A2A, and MCP focuses on discovery, identity, communication, and observability so agents from different vendors and clouds can cooperate on business tasks. But once these systems are in production, he says, collective behavior can fail even when individual agents seem to be working.
Pandey says the core problem is not just mistakes by single agents but compounding cognitive failures: misaligned intent, local optimization, and bad assumptions that spread through shared context. He claims that with just a few such failures, performance can collapse while dashboards still look green, because the system becomes internally inconsistent even as messages flow cleanly.
He also rejects the idea that enterprises can simply choose single-agent systems instead of multi-agent ones. In his view, real companies will always have multiple specialist systems collaborating, so the answer is better infrastructure, not fewer agents. That same logic extends to computing more broadly: he says determinism is giving way to probabilistic workloads, horizontal scaling matters more, and networking will have to evolve to support agents, embodied systems, and quantum nodes.
Key takeaways
- Pandey says connected agents are now the default enterprise model because companies will use systems from Salesforce, Cisco, Microsoft, ServiceNow, and others together.
- He says multi-agent systems can look healthy on dashboards while suffering severe cognitive failures that push results in the wrong direction.
- Pandey argues the real missing layer is shared intent, shared context, and collective reasoning, not just agent-to-agent connectivity.
- He says enterprises cannot avoid multi-agent collaboration because business workflows already span many specialized systems and teams.
- Pandey expects quantum networking and post-quantum security to become urgent, and he says Q-Day could arrive in 2029 or earlier.
Notable quotes
“connectivity was just the first layer. So, discovering these agents, plumbing them together, to then solve for a business outcome”
“100% of multi-agent systems had at least one cognitive failure. Misaligned intent, non-coordination, doing something that you could not verify towards a goal”
“the answer is not fewer agents. The answer is better machinery, better infrastructure that allows agents, just like human beings, to then collaborate as subject matter experts to solve for a business outcome”
How the episode opens
What will multi-agent systems look like when they go from training to production? Let's talk about it with Vijay Pande, the SVP and GM of Outshift by Cisco, in a conversation brought to you by Outshift by Cisco. Uh Vijay, great to see you. Thank you for being here.
It's a pleasure to be here, Alex.
So, uh Outshift by Cisco has been on top of this idea that agents are going to work together to get things done for a long time. We spoke with your colleague Guillaume a few months back talking a little bit about the Internet of Cognition and how uh your division within Cisco really has been doing the research and looking at how multi-agent systems will collaborate and get things done. Then a crazy thing happened, right?
Uh the world started to become exposed to the fact that agents can work together. Of course, when we saw what happened with OpenAI and Hugging Face hack, and it led to lots of questions about what this is going to look like when it goes towards productive uses uh as opposed to what we saw in the past, which was a destructive use. So, we have the perfect uh person in the right conversation here uh to go into what happens when agents work together because you've actually worked on doing this in practice.
So, I'd love to hear a little bit first about what you've built when it comes to multi-agent systems.
Alex, agents have gone through three phases pretty quickly. They've gone from isolated agents to islands of agents, and these are agents within a single harness. So, whether it's OpenAI harness or Anthropic harness or your harness of choice, to then moving towards connected agents. So, isolated islands to connected.
From the episode’s captions: roughly the first two minutes, lightly cleaned. Keep watching on YouTube →
Topics
- multi-agent systems
- agent observability and guardrails
- cognitive failures in agent teams
- scale-out computing
- post-quantum security
Companies and tickers mentioned
- CSCO
- Cisco
Watch the episode
Watch “How Agents Can Work Together — With Vijoy Pandey” on YouTube
Summary, takeaways and quote selection by Octagon. Quotes are short excerpts; the full conversation is in the episode.
More from Big Technology Podcast
-
AI Stockpocalypse, Anthropic Nears RSI?, Prediction Markets vs. Pollsters -
Will Jensen Huang's Affair Be The End of Nvidia? -
Can AI Keep Going After The Democrats’ Big Win? — With Teddy Schleifer -
Big Tech's Ugly Week, Getting Rich At Anthropic, AI Costs Plunge -
Google’s Potential, AI Money Spigot Shutting?, PHP Bug & Cybersecurity’s “Black October” -
How To Take Enterprise AI Pilots To Production — With Russell Jukes
Big Technology Podcast, this episode and its artwork belong to Alex Kantrowitz. Octagon is not affiliated with the show; this page is an independent summary.
