It was something of a surprise to learn that Sudheesh Nair, President of Nutanix, had suddenly jumped ship and moved to become the CEO of analytics-through-search vendor, ThoughtSpot. So it seemed only polite to ask him why, and more to the point, where he sees the company heading under his tenure.
His answer was interesting, coming from a man who seemed to have a clear vision of where Nutnaix was heading, why it was heading there and how the journey was progressing. Basically, he likes to feel uncomfortable, and felt the need to disrupt himself.
But there was also a side salad to this meal which suggested a sense of growing frustration with the infrastructure marketplace and the view of many C-level placeholders amongst the customer community. In particular Nair was finding the lack of vision amongst them, an unwillingness to have a sensible view of the possible rather than a fear of a perceived inevitable, a rather long road to travel::
At the top C-Level, when it came to infrastructure, the talk was always just about price. It was impossible to get them to talk about the possibilities for exploiting their infrastructure to make more money for the business. All they want to talk about is saving money.
When the opportunity arose to take over the helm at ThoughtSpot, therefore, he readily took on the job. After all, there was an element of staying within `the family’, as one of ThoughtSpot’s co-founders, Adjeet Singh, was also one of the founders of Nutanix. There was also the aspect that the company’s whole thesis – using search technologies as the primary tool for extracting more insight from any user’s available data – is aimed at the notion of helping them to make more money.
He also sees a challenge for himself in changing what he sees as misperceptions in the marketplace about ThoughtSpot:
Yes it comes across as search, and yes it is search-driven analytics, but the real issue is that search is the underpinning for how the modern internet works. Search is now the way that people enter the world of information, and often they don’t even think of what they are doing as `search’. Yet it is also extremely complex, especially if trying to search structured data without using SQL querying.
But what is of most interest to him is that it is also based on what he sees as a good AI. He sees this combination being a significant disrupter because the exponential rate of data growth means that any value-generating interaction will require what he refers to as some kind of ‘guard rail’.
Targeting the target
This requirement, however, does then point at the realisation of such a guard rail in pre-packaged, use-case focused services, either as software that can run on-premise or even as a SaaS-delivered resource, or as a combination hardware/software appliance form factor.
It is a direction that is a part of development plans that should see fruition by then end of this year. It won’t be an `appliance’ in the classic hardware/software package form factor, but it does form an important part of his wider thinking. He sees one of the important requirements being the ability to produce targeted results at ‘the speed of thought’ where, for example, typing in a search argument (even a simple one such as a name) may start with quite a broad church of possible answers. But any subsequent selection then quickly starts the process of refining the areas of search, so that the AI component can start pre-selecting evermore granular and refined suggestions and answers.
Nair feels that thinking of ThoughtSpot as an appliance is missing the point, not least because only 15% of its customers are on the cloud. For now at least the core marketplace is on-premise, large-scale analytics of data regardless of where it is held. Some of that data may well be held in cloud-based services, but the work is done on-prem, as the goal is making it easy to map business-related data – for example about customers – held on one application or service with data about the same customer held in other applications and core back-office databases. He says:
It is going to be a multi-cloud world; the data will be living in the data center in Oracle databases and other data warehouses. Users will have SAS applications like Salesforce and others, they will have specific applications that they have built. Now, the question is, can they bring it all together from a context point of view? So now, you need to have an in-memory database that is stretchable and elastic, because in the cloud it has to be elastic. It needs to stretch to on-prem, which breaks pretty much every basic computing model out there.
According to Nair the company has some customers running billion-row queries, and are getting answers of value as they are typing. He can see some validity in a comparison with IBM’s Watson, though he also sees a gap in execution that comes from the heritage of ThoughtSpot, where the early team members came from Google and brought with them experience of infrastructure scalability in memory, performance and elasticity, onto which the AI component was overlaid.
Hitting the spot
The introduction last year of SpotIQ, as an add-on to ThoughtSpot is now being used as the basis of some new developments that will allow targeted implementations of the system to be created. This should provide the opportunity for users, ThoughtSpot itself and, quite possibly, some members of the channel partner community to introduced tailored versions that focus on the search and analytics needs of specific market sectors.
SpotIQ has added such functions as automated and personalised insight generation, a natural language interface and the ability to create customised and scheduled analyses. The latest version, which takes the SpotIQ capabilities into the market for targeted analytics tools, is currently in Beta test and is expected to be made available at the company’s `Beyond’ conference in December. The aim is to help users exploit and monetize the exponential growth in data which leads to one of the core problems of big data analytics – the search for that needle in the ever-growing haystack. Nair observes:
The haystack is becoming so massive the question becomes how to figure out where one be looking for the needle. So in a sense, you need the system to say, look, there’s a higher probability that the needles hiding on the left most corner of this stack, or somewhere on the right side of that one, because it is difficult to figure out the right question to ask when there is so much to deal with. You need the system to tell you the questions you should be asking. So, AI-driven analytics is moving from giving answers to your question to telling you these are the questions that you should be asking.
The goal here is to help the users rapidly reach that point where the AI-guided question connects with the experience and knowledge – both formal and tacit – that the human user has about the business, the marketplace, the product or technology, and the specific customer so that the most intuitive questions, more intuitive than the system could devise, emerge. But the process depends on using the AI to rapidly funnel them in the direction that they need to go.
This also means that, particularly for the classic business analytics applications such as analysing weekly sales results, there is no need to hire expensive and rare data scientists for, according to Nair, everyone can directly interact with the data as they are typing their questions.
Combining search technologies with AI as the basis for business data analysis tasks, looks like a useful approach, particularly for those tasks that form the stock-in-trade start point for many business decisions, the ones that start with questions such as: `what is happening, and why?’
So it will be interesting to see where Sudheesh Nair takes the combination as he beds in as the new CEO of ThoughtSpot. The potential to pre-configure the system to the needs of specific market sectors should open up new markets for the company, especially where experienced and intelligent staff can achieve as much, or more, as analysis tools that are best used with attendant data scientists.