What We Do

What is Smesh

Smesh makes beautiful things powered by real-time data, usually from social networks. Sometimes consumer-facing, sometimes for behind-the-scenes intelligence.

  • The ‘Film Pulse‘, a real-time film-ranking system for Orange, powered by tweets, Facebook activity and blog content.
  • Extending the capabilities of Zeebox to enhance its social data capabilities.
  • Measuring attitudes in cities around the world for Diesel Jeans.
  • Social data ‘fuel’ from Facebook and Twitter to power a physical animatronic robot for Orange Phone Fund.
  • A human-friendly ‘tagbot‘ to provide live access to the Guardian’s content through Twitter.
  • Many more examples in our case studies

Smesh has a core layer of technology components (including many Open Source tools) that work together to deliver much of the functionality needed for our bespoke projects; we add on custom layers and enhancements as needed – for example, we developed our own sentiment analysis system for the Film Pulse project.

This enables us to deliver high-end, high-performance social data products quickly and effectively, using battle-proven systems.

How Smesh is different

Smesh sits a layer above most of the other things we get asked this question about. For example, social data API providers like DataSift and Gnip, and social monitoring tools that have APIs for data (Brandwatch, Radian6, Sysomos, and many more) are all things that Smesh can in principle consume data from and work with.

For most things we build, having this source data (even if it comes with lovely meta-data like Klout scores attached) is just a starting point. The data then needs custom processing and sometimes cross-referencing with other bodies of data or archived information.

Real-world features

Smesh has been used for many high-performance campaigns, so sports many essential large-scale campaign features:

  • Multi-language, multi-territory Content Management System
  • High-availability deployment for maximum uptime
  • Load-balancing and multiple caching layers to support thousands of API calls or page views per second
  • Flexible moderation options to support varying degrees of client and legal appetite for risk, including full pre-moderation, post-moderation and automated moderation
  • Automated language detection
  • Automated profanity filtering