This morning I met with Charles Nicholls, CEO of SeeWhy. It was an accidental encounter but one that was well worth the time. Charles used to be with BusinessObjects and is a long time veteran of the BI space. He knows his stuff. And while it's a few years since we last met, I've always found Charles to be a person who deals in reality and not faction (fact as fiction). Today was no exception.
SeeWhy is solving interesting problems that require real-time actions from the analysis of streaming data. For instance – real-time discovery of events impacting retail sales or the likelihood a web transaction is a fraud. Some – like the web verification example – need to be automated. But I see deep value for 'act on data in motion' applications in those areas where human intervention is required to tackle exceptional business events. It's what I like to term 'process-light intelligent action.'
Exceptions are, by their nature, outside the process. Why therefore would you rigidly impose process on unexpected and usually unpredictable events, even when required action could be expressed as part of a wider process? The point is that in tackling a thin slice of activity – say tracking a particular promotion in a specific store(s) – you might be able to avoid stockouts by changing the replenishment process to allow for in-store rather than next day for the promotional period.
In my retail example, combining SeeWhy analytics with flexible process provides a reason for capturing POS data in real-time, something which is usually reserved for managing predicted demand rather than actual. Taking this approach plugs the gap between assumed knowledge and on the ground reality.
SeeWhy can deliver information to email, SMS and, because it spits out XML, it can support RSS feeds though that specific functionality has not been built. It can also push information to SAP systems where there is a requirement to trigger processes inside SAP. Alternatively, it can take and supply data to mioddleware systems like webMethods and TIBCO.
SeeWhy uses pattern recognition techniques as the foundation for creating views of aggregated information. It does not rely on creating SQL queries so doesn't exhibit the processing load characteristics of a typical data warehouse. Although it looks 'lightweight' there is a lot of heavy lifting going on in thre background.
Today, SeeWhy has limited traction – its flagship customer is Diageo's US Guinness shipping business and has other clients who are working on fraud detection projects. It has partnered with Accenture to deliver custom solutions.
Pricing is on a per processor basis with annual fees representing licence, maintenance and support.
There's a lot more to this story – I'll be revisiting it in time.