YouTube now allows users to build
custom feeds by allowing them to describe the types of videos they want to see or recommend.
Think of it as letting users build their own algorithm. The company launched a "custom feed"
feature on Wednesday.
Creating a feed means typing the description into a prompt box on YouTube, whether that is a feed dedicated to a particular interest or one that serves another
purpose.
For instance, viewers can build a custom feed around video podcasts or ask for videos to go with a particular mood.
The prompt feature relies on Google Gemini AI model to
build the feed around the request, and the resulting feed is then pinned to the top of the user's YouTube home page in its own tab.
AI-powered requests can be described in length, such as
details about how the user may want to curate the feed, what it should include or exclude, and what content to prioritize.
advertisement
advertisement
The company also launched new capabilities to
"Ask YouTube" like asking multiple questions as needed through an interactive experience.
When searching for reviews of a specific product, users may receive organized
video recommendations, such as a comparison table that includes product attributes and categories based on your preferences.
YouTube made the announcements at its annual Made on YouTube event
today. This requires both network capacity and speed.
The connection to the data Cisco Systems released today involves intent-driven
computing. YouTube’s custom feeds are an example of user-facing agentic behavior that causes massive online traffic spikes Cisco predicts.
Cisco released data today that shows just how much capacity and how much bandwidth are required based on AI adoption
rates.
The data comes from survey results released today of more than 1,000 IT and network professionals. The data shows that 75% have already deployed AI for network operations that
combines AI and machine learning, with more than half now running agentic AI systems. About 84% expect to run an AI-led operating model within 12 months.
Operational complexity is compounded as AI adoption grows and agents operate at software speed, generating unpredictable traffic patterns and greater data exchange, according to
Cisco's report.
Based on the data, Cisco found that AI traffic is on pace to double every six months.
When traffic generated by agentic AI is included, Cisco testing found, tasks
performed by agents can generate up to 450% more total network traffic.
Companies are increasingly turning to AI and agent-driven networks. Many organizations are allowing AI agents to take on
more responsibility with network operations -- including making production changes such as rerouting traffic, adjusting wireless parameters, isolating suspicious endpoints, and resolving incidents end
to end without prior human approval.
Some 75% of companies participating in Cisco's report have already deployed AI for network operations, while 51% run agentic AI that performs actions today
and 80% are comfortable granting AI a high or fully autonomous role, including 24% who are comfortable with AI acting with no human oversight.
Interestingly, 82% are comfortable with allowing
AI to make at least some production network changes without prior human approval, while 84% expect an AI-led operating model within twelve months.