Some weeks back I got an update from Lyzasoft but I have not been able to post it until now. I first wrote about Lyzasoft back in 2009 and since then the company has been busy. The company’s vision remains the same – to create a way for end users to work with BI that is collaborative, providing integration, enrichment, manipulation, visualization, calculation, distribution and sharing in an integrated environment. They see this as cutting across lots of traditional market segments as the way people actually work does not fit these segment definitions.
The current Lyza Studio product (available for Mac, Linux, Windows) lets you connect to databases on remote hosts and do local computing manipulating the data. The workflow for these data manipulations is all federated, running on your local laptop. The product allows some nice instant building of graphs by dimensions and then allows users to drill in. It supports integrating databases, allowing automatic re-aggregation and requires no pre-defined drill down paths etc. They have added more chart options such as a horizon chart (standard deviations from the mean on a daily basis) and segments can be broken out for parallel charts, each normalized based on their own volume (correcting for large volumes in one segment or instance). Easy aggregates like moving average, sum to date etc can be quickly add to the chart to provide some lightweight statistics. As before I really like the way the trace lineage can be easily found – shows the steps used to produce a chart or a set of data. If one is trying to manage manual decision making then this gives you something that is analogous to the way rule logs help show what automated decisions were made.
Beyond the Studio product they have been upgrading the web client and the associated Lyza Commons hub. You can publish to the Lyza Commons hub from the Lyza Studio. The Lyza Commons hub has different kinds of artifacts like graphs, data collections or bookmarks being shared. Everything displayed is filtered through “trust settings” so that data is only shared with the folks who are identified by the author as trusted. Groups can be created by an individual or managed by the corporation for the community as a whole based on LDAP/ActiveDirectory etc.
Using the web client you can write an article (blog) about the decision(s) you made based on the data. You can embed a feed into the article and Lyza snapshots the data and the process of creating that data at this point. This creates a version so that even if the workbook and the data changes, the article still has the version on which it was based. Articles are written using the WordPress environment (open source blogging software) with addition functionality to make it easy to link to the various exhibits or artifacts within Lyza. A reader has the option to see the current version of the data and/or workbook and the “snapshot” is active in that it can be explored and developed. Articles can also include explicit links to the source XML of the trace. Thanks to this you have access to the data behind someone’s graphs or reports.
You can tag, bookmark, rate and comment on people’s items using a familiar social media-like interface. Users can also subscribe in an RSS like way to be shown those things that they care about but have not looked at since they were last updated. Scheduled updates are supported and searches include comment streams as well as titles, descriptions etc. Comments also show in a twitter-like feed. All these social/sharing capabilities use familiar metaphors and should help engage more users in their data and what it means.
The underlying XML model is bound to everything it produces – so if a user publishes a graph, for instance, then the XML that describes how to get that graph is linked to it. Publishing these XML models to Lyza creates a hub that allows people to interact with the chart, blog about, create derivatives of it etc. Each of these would be bound to the XML. Users don’t have to use Lyza Studio either, they can use the web environment and any changes or derivatives made maintain this XML path.
Even on the web the graphs are not pictures but are interactive, allowing people to engage with the data. Users can add filters, add new metrics, derive new data – everything that you could do in Lyza Studio. The difference is that the web client does not allow the same construction of complex multi-step workbooks quickly – everything must be done one step at a time. This is fine for editing but would be restrictive when developing new models (hence the availability of the Studio product).
This separate web client/hub model also allows a federation server to support a collaboration environment (based on the web client) while allowing corporate databases to stay “behind” Lyza Studio. Manipulations that derive new data from the same feed can be built in this environment without creating new data. You end up with lots of sibling models – different perspectives all sharing the same data feed.
Today Lyzasoft has very focused deployments – departmental –and this makes the features for faceted search , popular search terms and graphs etc very powerful. I think they will need something more than “most popular” in situations that are supporting multiple disparate groups but their current implementation is interesting and will help to generate engagement with the data.