AI's next bottleneck isn't the models — it's whether agents can think together

AI's next bottleneck isn't the models — it's whether agents can think together

AI's Next Bottleneck: Can Agents Think Together?

Section 1 – What happened?

Vijoy Pandey, Senior Vice President and General Manager at Outshift by Cisco, has highlighted a significant challenge in the development of next-generation AI systems. According to Pandey, current AI agents can be connected and stitched together in workflows, but they lack the ability to think together. This limitation stems from the absence of semantic alignment and shared context, forcing agents to work from scratch each time they interact. Pandey emphasized that connection is not cognition and that achieving shared cognition is crucial for unlocking the full potential of AI.

Section 2 – Background & Context

The concept of shared cognition refers to the ability of AI agents or entities to work together to solve complex problems without human intervention. Pandey drew an analogy between human intelligence and AI, pointing out that humans evolved from individual intelligence to collective intelligence through improved communication. Similarly, Pandey believes that AI systems need to reach a point where they can share cognition, enabling collective innovation, shared intent, and coordination. This requires a new infrastructure, which Pandey calls the "internet of cognition."

Section 3 – Impact on Swiss SMEs & Finance

The development of shared cognition in AI systems has significant implications for businesses, including those in the Swiss finance sector. As AI becomes more integrated into workflows, the ability of agents to think together will be crucial for solving complex financial problems, such as risk management and portfolio optimization. Swiss SMEs, in particular, may benefit from the development of shared cognition, as it could enable them to tap into the collective intelligence of AI systems, improving their competitiveness and decision-making capabilities.

Section 4 – What to Watch

The next step in AI development will be the creation of new protocols to support next-generation agent communication. Pandey's team at Outshift by Cisco is working on a set of protocols, including the Semantic State Transfer Protocol (SSTP), Latent Space Transfer Protocol (LSTP), and Compressed State Transfer Protocol (CSTP). These protocols aim to codify intent, context, and collective innovation as a set of rules, APIs, and capabilities within the infrastructure itself. As these protocols emerge, they will likely have a significant impact on the development of AI systems and their applications in various industries, including finance.

Source

Original Article: AI's next bottleneck isn't the models — it's whether agents can think together

Published: April 15, 2026

Author: taryn.plumb@venturebeat.com (Taryn Plumb)

---

Disclaimer: This article is for informational purposes only and does not constitute financial advice. Consult a licensed financial advisor before making investment decisions.