Artificial intelligence is getting a major upgrade on two fronts: how companies talk to you, and how the supercomputers training these systems talk to each other. On the customer side, an AI voice company called Ringg just raised $10 million in new funding. This brings its total early-stage funding to $15.5 million. Ringg already processes 20 million call attempts a month. Instead of making simple robocalls, these AI agents handle complex tasks. For example, they can book doctor appointments or help you buy items you left in your online shopping cart. Ringg plans to partner with offshore support hubs to bring these smart assistants to companies in the US and beyond.
Why this matters to you: You will soon talk to much more capable AI assistants when you call customer support, and the systems powering them will run faster and cheaper.
To build these smart tools, tech companies must connect thousands of powerful computer chips. Meta recently introduced a new networking technology called MetaRoCE to help these chips share data much faster. Right now, training AI is like a group project where everyone has to wait for the slowest person. If one packet of data gets lost in the network, the whole system slows down. MetaRoCE fixes this by letting data arrive in any order. This means the computer chips never have to stop and wait.
Meta currently uses hundreds of thousands of computer chips to train its AI. Because these networks are so massive, even tiny delays can waste a lot of expensive computing power. Meta plans to release the official blueprints for this new networking system by October 2026. As both the physical networks and the friendly voices of AI improve, our daily interactions with technology will only get smoother.
