Moderation couldn't keep pace with 600M users. AI agents now filter toxicity while models recognize 2.5B objects to refine search.
A visual search and discovery platform serving 600 million monthly active users and managing over 500 petabytes of data.
Scaling to hundreds of millions of users created immense pressure to maintain a safe and inspirational environment. Traditional engagement algorithms...
“For over a decade, we've been leveraging AI to craft a uniquely positive online experience, striving to make every moment on Pinterest additive, not addictive.”
Visual discovery engine for recipes, home inspiration, and style ideas.
Cloud computing platform and on-demand infrastructure services.
Pinterest's Visual search and discovery is part of this use case:
Related implementations across industries and use cases
Designers manually browsed thousands of files to find assets. Now, AI lets them locate components via text or screenshots.
Each team built its own stack; every request reprocessed thousands of images. A shared platform with precomputed embeddings cut latency 44x.
Raw pace and power data overwhelmed users. AI now turns millions of daily uploads into instant, encouraging performance insights.
Each team built its own stack; every request reprocessed thousands of images. A shared platform with precomputed embeddings cut latency 44x.
Elasticsearch treated creator vectors like log data, aging 400M+ to frozen storage, pushing search to 25s. A vector-native store fixed it.
Scattered data and basic coding tools bottlenecked engineers. A 9-agent AI workflow shifts them from writing code to directing AI teams.
Sequential AI testing bottlenecked development. Engineers built a concurrent, code-first pipeline to evaluate agent responses in seconds.
Tournaments running simultaneously meant an hour of manual checks each. AI agents now run them in minutes, freeing the team to be proactive.
Large AI training jobs meant fighting for preemptible slots or leaving campus. Marlowe gave any lab guaranteed multi-node access on demand.
Moderation couldn't keep pace with 600M users. AI agents now filter toxicity while models recognize 2.5B objects to refine search.
A visual search and discovery platform serving 600 million monthly active users and managing over 500 petabytes of data.
Scaling to hundreds of millions of users created immense pressure to maintain a safe and inspirational environment. Traditional engagement algorithms...
“For over a decade, we've been leveraging AI to craft a uniquely positive online experience, striving to make every moment on Pinterest additive, not addictive.”
Visual discovery engine for recipes, home inspiration, and style ideas.
Cloud computing platform and on-demand infrastructure services.
Pinterest's Visual search and discovery is part of this use case:
Related implementations across industries and use cases
Designers manually browsed thousands of files to find assets. Now, AI lets them locate components via text or screenshots.
Each team built its own stack; every request reprocessed thousands of images. A shared platform with precomputed embeddings cut latency 44x.
Raw pace and power data overwhelmed users. AI now turns millions of daily uploads into instant, encouraging performance insights.
Each team built its own stack; every request reprocessed thousands of images. A shared platform with precomputed embeddings cut latency 44x.
Elasticsearch treated creator vectors like log data, aging 400M+ to frozen storage, pushing search to 25s. A vector-native store fixed it.
Scattered data and basic coding tools bottlenecked engineers. A 9-agent AI workflow shifts them from writing code to directing AI teams.
Sequential AI testing bottlenecked development. Engineers built a concurrent, code-first pipeline to evaluate agent responses in seconds.
Tournaments running simultaneously meant an hour of manual checks each. AI agents now run them in minutes, freeing the team to be proactive.
Large AI training jobs meant fighting for preemptible slots or leaving campus. Marlowe gave any lab guaranteed multi-node access on demand.