- The Bot-2-Bot paradigm quickly becomes an agent-to-agent
- Network Supremo Cisco is at the forefront of this evolutionary bending point
- Agents will probably also have their own language and bypassing human lingo
The next boundary of artificial intelligence unfolds, and worrying for many, human involvement may not be a key requirement.
Discussions in tech circles are increasingly focusing on building a framework where AI agents – autonomous, reasoning machines – collaborate seamlessly to achieve complex tasks independently of human supervision. Baptized “Internet of Agents”, this next step could eventually redefine how companies and industries work.
Cisco’s Outshift group is one of the most important players who tackle the challenges of creating the infrastructure needed to connect these AI agents.
DroidSpeak
The vision is certainly ambitious. AI agents from different companies and organizations must be able to automatically find each other, exchange data safely and solve problems in collaboration.
“We are standing at another bending point – the move to distributed agent computing,” Outshift’s ViJoy Pandey noticed in a recent white book. He described this as “a platform paradigm shift larger than all other platforms in the past.”
Pandey claims that this Internet of agents must be “fast, secure and open” containing advanced hardware software stacks for low latency and quantum-safe encryption communication to ensure interactions.
The emergence of AI-driven “Foundation models” as Chatgpt has already accelerated the development of AI agents that go beyond generating text or images for planning, reasoning and working autonomously. Of course, there are still significant obstacles. Industrial cooperation is necessary to establish ordinary protocols, ensure system scalibility and build trust frameworks.
Microsoft scientists take another yet complementary approach by tackling communication bottlenecks between AI agents. Singularityhub Reports The Techngiant has developed “DROIDSPEAK” Star Wars. It is designed to be faster and more effective than human language as English.
In a paper that is published on Arxiv Preprint Server, Microsoft says, “Experiments on different data sets and model pairs show that DroidSpeak achieves up to 3x higher flow and 2.6x faster prefill times with insignificant accuracy loss compared to full recupation.”
Despite the promise, both Cisco and Microsoft recognize that the road in the future is long. Open standards for cross -compatibility between models and progress in calculation efficiency will be necessary to fully realize an internet of agents. But if it succeeds and there is no reason to suspect that it will not be, the concept can create a new era with collaboration and one where people play only a peripheral role.
Via The next platform
