Connexica CXAIR

CXAIR from Connexica (which was previously Ardentia Search) is a BI tool specifically designed to support queries that span both structured and unstructured data, though it can equally well handle either one or the other on its own. It uses indexing (as in search-based indexing not database indexing) to extract structure from text, spreadsheets and so on and to support queries against that structure.

For relational data it treats each row in a database as if it was a document and, again, creates relevant indexes to support query capability. These indexes are supported by the use of double Metaphone and Soundex algorithms, as well as Levenshtein distance metrics, to enable indexing against fuzzy matches. Indexes can be updated incrementally or completely, as required.

In most respects the way that you actually use CXAIR is pretty much like any other BI tool: you have drill-down, tables, lots of graphic options, integration with Google Maps (so that you can have a heat map, for example, superimposed over a map) and so on. Plus, of course, there is the big advantage that inexperienced users can simply use a search-like interface and don’t require any training. As an instance of this, one of Connexica’s partners (see later) has replaced Crystal Reports as its embedded BI and reporting tool because CXAIR, at least in part, is so much easier to use.

The other big difference from conventional tools is that the product is set-based: you create a set by selecting filter options and then you can manipulate the results in a variety of interactive ways including Venn diagrams. Unlike other tools that have limited support for Venn diagrams (often two sets or subsets only), with CXAIR you can have as many as you like and you can drag and drop them to intersect one another and the numbers in each sector automatically update, which is very useful if you want to examine relationships or overlaps of product sales, segmentation classification and so forth. You can also change the shape of the “circles”, re-dimension them at all will, and so on. In effect, what CXAIR is providing is set-level analysis that supports joins, unions, intersections, correlated sub-queries and so on.

The only other BI product that I have seen that supports this deep level of set processing is/was SAP Business Object’s Set Analyzer, which was a great product but never sold well, partly because it didn’t integrate tightly with the rest of the Business Objects products range (it was acquired from its original developer) and partly because it was far too expensive. However, it had a great concept that Connexica has now realised through the application of search technology.

In terms of its go-to-market strategy Connexica (which raised a first round of VC funding last year—it was originally self-funded) uses a mixed channel and direct marketing based approach though its major emphasis is on third party companies embedding its software within their own applications (in sectors such as healthcare and capital markets). For direct sales the company focuses on marketing companies (who want to do things like sentiment analysis) and retail organisations. Connexica is also growing its direct partner (resellers and distributors) base and through these, and by its own efforts, it has a number of installations in the United States, and a growing customer base in various European countries and the Middle East.

Perhaps the most important thing about CXAIR is that it looks like the right tool at the right time. The company has been beating the drum of easy-to-use BI (much, much easier than conventional BI tools) for some time. And also the ability to combine structured and unstructured (hybrid) data into a single query. But now demand for the latter has taken off with the advent of “big data”.

Unlike all the big boys in the BI space, CXAIR was designed from the ground up to support queries against hybrid data while the best that its competitors can do is to bolt separate products together and pretend that they are integrated. Connexica looks especially well positioned.