Technology vendors hold press conferences. They like them, it’s what they do.
But over and above the formulaic press conference process and experience, some technology vendors also stage conferences, exhibitions and annual symposia. Again part of the standard arsenal of customer, user, partner and press interaction, these are tried and tested tools.
Despite the degrading effects of the pandemic, most of these engagement scenarios still exist. Well, why not? How else are we going to be able to hear the CEO repeat themselves and regurgitate their keynote highlights an hour later in front of assembled press and analysts?
There is, however, a middle way.
Tech study tours
Not a formalised industry term and a long way from becoming any kind of de facto piece of terminology that the corporate marcoms team might jump on, a small number of organisations have aimed to engage with the press by saying something like: there’s no news, just come and meet our software engineering lead, talk to our chief architect, shake hands with the CEO and ask them a few questions and eat sandwiches in our staff canteen to get to know us.
ThoughtSpot is one of those companies.
Always keen to dive into software engineering constructs that emanate and stem from data science roots, the Computer Weekly Developer Network team embarked upon the ThoughtSpot ‘study tour’ 2023 this month in Mountain View, California, in order to get our hands dirty in the mechanics of the company’s Modern Analytics Cloud service and find out why it now calls itself the AI-powered analytics company.
Below follows a number of comments drawn from meetings carried out in situ this month as we posed the question, what’s next for AI analytics?
According to ThoughtSpot CEO Sudheesh Nair, we (as a whole in the IT industry and perhaps as humans at large) are at another key inflexion point in technology with the arrival of OpenAI and the new generation of generative AI models.
Accessibility vs. expressibility
“It might be an Internet Explorer 3 moment, an iPhone moment, an AWS EC2 moment or some other seminal moment in time for the industry, but in terms of data science and data analytics, we’re at an important crossroads,” Nair. “You can think of it as the difference between accessibility vs. expressibility – what I mean is, a businessperson might have many great ideas for an application or service, but they know they’re unable to express them in C++ – so generative AI democratises expressibility in that sense. People are often afraid of data and are much more comfortable with opinions, when we give these users the chance to query data analytics in natural language, then we achieve that special moment… and that’s what’s happening right now.”
CEO Nair is sanguine about all the opportunities being thrown up in this space. He thinks that with all the Large Language Models (LLMs) being so impactful and important to current developments, the smart money is on the combined use of Small Langauge Models (SLMs) – a scenario where the dataset analyzed is proprietary and kept within the confines of any single organization – in order to get the strongest hybrid mix and perhaps best of both worlds.
Right place, right time?
Speaking to Amit Prakash, ThoughtSpot co-founder and CTO, there’s clearly a mutual understanding of what’s happening here.
Asking CTO Prakash whether ThoughtSpot is – happily perhaps – in the right place at the right time, he reminds us that the company has been working in the data analytics realm for 11 years now. His own history traces back to being on the team that built Microsoft Bing and also working on the probability model science team at Google AdSense, so he has seen the development of algorithmic improvement being championed at the core.
“When we founded ThoughtSpot, we had a very different approach to analytics… we were at the point where we needed to move away the now archaic use of dashboards towards Google-like search,” said Prakash. “The problem with dashboards is that somebody thought of the questions [you can ask of them] six months ago and you’re stuck with those as your basic options. ThoughtSpot has gone beyond dashboards to create what we call Liveboards that have the ability to be constantly queried and questioned.”
Queries are defined by the data model that the data analyst builds at the start – and these can be augmented, extended and connected – which does occur in some use cases. Overall it’s a lot more freedom and power. It’s still possible to edit in SQL if a user feels they need to engineer queries in some other nuanced way that is not initially possible.
Let’s remember what technologist Alan Kay said: “Simple things should be simple, complex things should be possible.”
Moving to talk with Sumeet Arora, chief development officer and Bhargav Addala, SVP of product management at ThoughtSpot – it’s all about the birth today of what is being called a ‘modern data experience’. We are going beyond what was initially built by data analysts for domain specialists only. Beyond this point, we are now able to provide data analytics to all knowledge workers.
Because there is the public availability of AI that enables us to interpret Natural Language through the use of LLMs. What Arora and Addala are working on is taking that ability and connecting it to the data analytics function in ThoughtSpot as a platform in and of itself. As we now move to an era where this kind of data analytics is not only available for mobile devices, but also optimised for mobile devices, users are consumption-empowered to get analytics wherever they need.
Deep into the delight-factor
According to Cindi Howson, chief data strategy officer at ThoughtSpot. Businesses can not take a ‘wait and see’ approach in terms of implementing new generative AI and the new era of cloud. She suggests that any firm that thinks they need to just focus on keeping the lights on is no longer good enough.
“It’s all about needing to realise that we need to delight customers and give them a new kind of experience across all products and services. That delight factor is not just a marketing term, it’s all about meeting customer needs and expectations at every level… this is all because the ‘switching costs’ [abandoning an online cart and moving to another supplier] in the digital economy are so low,” said Howson.
She further suggests that when we now move forwards with personalisation and hyper-segmentation [a wider set of demographics to define any given customer] so that we can really detail the needs of customers down to the most granular level, the new data-driven digital business economy really starts to flourish.
We wrapped up this study tour with Kuntal Vahalia, SVP of worldwide channels and alliances at ThoughtSpot. Given that ThoughtSpot is not a full-stack product as such and exists as the experience layer in the modern stack – with vendors such as Snowflake, AWS RedShift, Databricks, Google Cloud Platform, and Microsoft Azure sat below… this is what Vahalia looks after.
The relationship with these firms is very much a two-way push-pull interchange in terms of how it affects the ThoughtSpot innovation roadmap. In terms of all those alliances, Vahalia always looks for ‘intention and focus’ as key factors in terms of their involvement in the whole customer lifecycle process. He says that every union the company picks is a deep joint product partnership that is extremely go-to-market focused.
“Some organisations do a great job of building great products together, then they forget how to execute deep distribution level programs designed to get products (and services) to market successfully in a total enterprise product universe where hundreds and in some cases thousand are all vying for share of voice and share of market,” said Vahalia.
To sum up then, this was not a press conference, this was an opportunity to hear from real world software engineers working in their own space and behind as upfront and honest via non-scripted interaction and interrogation.
For study tours in general then, is it a case of we don’t need no education? Not at all, leave those kids alone.