Sisense adds embeddable chatbot to generative AI toolkit
The embedded BI specialist introduced Analytics Chatbot, a tool that enables natural language interactions with data, and a feature that lets users easily create summaries.
Sisense on Tuesday unveiled an embeddable generative AI chatbot and natural language narration capabilities that will enable customers to query data and ask follow-up questions to gain insight without having to write code.
Sisense Analytics Chatbot and the accompanying narrative storytelling capabilities are now in beta testing. Public preview is expected within the next few months, with general availability by the end of the year, according to Ayala Michelson, Sisense's chief product and technology officer.
Sisense already offers a natural language processing (NLP) tool called Simply Ask. Simply Ask, however, is a traditional NLP tool rather than a generative AI-powered capability. It doesn't enable the freeform natural language interactions or depth of analysis enabled by Analytics Chatbot.
Instead, the new capabilities, much like the generative AI assistants being built by Sisense peers such as Tableau and Qlik, enable users to query data using true natural language rather than business-specific language and receive data visualizations in response. In addition, the tool, fueled by the vendor's semantic knowledge graph, suggests follow-up questions that lead to deeper exploration and analysis.
Beyond Tableau and Qlik, AWS, Microsoft and MicroStrategy are also Sisense competitors that have already unveiled – and, in some cases, made generally available -- their own versions of an AI assistant. However, despite being later to introduce a generative AI-powered assistant than many of its competitors, Sisense is providing valuable capabilities, according to Mike Leone, an analyst at TechTarget's Enterprise Strategy Group.
"In the fast-paced world we live in, it may seem as though they're late. But it's important to not mix up speed with practicality," he said. "Sisense recognizes where their customers are in their journeys. And for me, the marriage of existing AI capabilities with GenAI and [large language models] drives the value here."
For example, the use of Sisense's semantic layer in conjunction with generative AI capabilities is significant, Leone continued.
"This is important to enabling the right type of data enrichment for an LLM," he said.
Based in New York City, Sisense is an embedded analytics specialist whose Compose SDK for Fusion platform enables customers to embed data products within the workflows of business users to simplify access to insights and help broaden the use of analytics within organizations.
The vendor's longtime CEO Amir Orad stepped down in April 2023. Ariel Katz, who had been Sisense's chief products and technology officer, was named his replacement. Following the CEO change, Sisense, like some other technology vendors as they adjust to post-pandemic conditions, began cutting staff, with reportedly laying off about 100 workers in July 2023 and another 60 in January 2024.
Domo and Logi Analytics, now part of InsightSoftware, similarly specialize in embedded analytics.
Enterprises might receive these seven benefits when using generative AI.
New capabilities
Generative AI has been the most dominant trend in data management and analytics over the 19 months since OpenAI launched ChatGPT, which marked significant improvement in generative AI capabilities. Even other trends such as increased use of vector search and a heightened emphasis on data quality are related to how they can enable enterprises to develop generative AI models and applications.
The simple reason is that generative AI now has the potential to make data management and analytics more than a specialized skill.
Data management and analytics platforms are complicated to use, requiring coding knowledge to integrate, prepare and query data as well as develop data products such as reports, dashboards, models and applications. In addition, users need to be data literate to interpret data.
As a result, studies have found that only about a quarter of employees within organizations have the needed skills to work with data as part of their job.
True NLP changes that, reducing the need to write code and lessening the need for data literacy training. LLMs, which have vocabularies as extensive as a dictionary and can infer meaning, enable that true NLP for the first time.
When NLP is combined with an organization's proprietary data, it minimizes the complexity of data management and analytics. In addition, it makes data experts more efficient by reducing time-consuming tasks.
Vendors have responded enthusiastically, integrating with LLMs to enable customers to develop their own generative AI tools as well as developing generative AI-powered tools on their own that their customers can adopt.
Sisense, which was one of the first analytics vendors to integrate with an LLM, is now joining the competition.
Sisense, however, is doing so in a different way than many of its contemporaries, according to David Menninger, an analyst at ISG's Ventana Research.
Unlike generative AI assistants that require users to be within a vendor's BI environment to make use of their capabilities, Analytics Chatbot and its accompanying narrative storytelling capabilities are now part of the Compose SDK for Fusion and can be embedded in the applications where employees do most of their work.
As such, the new features join two of the main ways enterprises attempt to extend analytics to more than just a team of data experts.
"These capabilities … allow Sisense customers to reach a broader audience," Menninger said. "There are two primary ways to [reach that broader audience]: embedded analytics and NLP. This offering addresses both."
Among competitors, MicroStrategy also offers an embeddable generative AI assistant. But given that Sisense is one of only a few vendors that specializes in embedded analytics, the embeddable nature of Analytics Chatbot helps it stand out, Menninger continued.
These capabilities … allow Sisense customers to reach a broader audience. There are two primary ways to [reach that broader audience]: embedded analytics and NLP. This offering addresses both.
David MenningerAnalyst, ISG's Ventana Research
"That's really their differentiation," he said. "Other vendors also go after this market, but Sisense focuses more intensely on it than others."
Once Analytics Chatbot is available, developers will be able to easily integrate the tool with business applications using multiple LLMs in conjunction with their proprietary data. Business users subsequently will be able to ask questions of their data in natural language to discover insights.
The narrative capabilities, meanwhile, will enable customers to create summaries and explanations of charts, graphs and dashboards using natural language.
The combination of multiple LLMs, Sisense's knowledge graph and proprietary data is noteworthy, according to Leone. The accuracy of generative AI outputs -- or lack thereof -- has been problematic. The inclusion of the semantic layer, which aims to ensure data quality and consistency, is therefore significant.
"Everything Sisense is doing with their semantic layer integration is particularly valuable," Leone said. "They are enabling deeper levels of trust by helping improve accuracy, reduce hallucinations and ensure far greater query execution."
Enabling trust is why Sisense took longer than many of its competitors to introduce a generative AI-powered assistant, according to Michelson.
Some vendors unveiled similar capabilities over a year ago. Some of those capabilities, however, are still not generally available. Instead, Sisense is in a group that includes Tableau and Qlik that have only recently unveiled their generative AI assistants.
"It's easy to take the basic LLM that was out there, play around with it and ask basic questions," Michelson said. "We wanted to introduce an offering that allows for accuracy, deeper types of questions beyond just descriptive analytics. … Narrations that are also part of the experience."
Plans
With the Analytics Chatbot and accompanying narrative storytelling capabilities now in beta testing, Sisense's roadmap is focused on a few core areas, according to Michelson.
One is providing customers with easy-to-use capabilities for data modeling so both data experts and business users can create the data models that foster analysis.
"We'll leverage a lot GenAI to accelerate that," Michelson said.
Included will be automated modeling capabilities, natural language generation features and explainability tools.
In addition, Sisense plans to add generative AI capabilities that will help surface insights, enable deeper analysis through follow-up questions and provide recommendations as customers work with data.
Menninger, meanwhile, said that although it's important for vendors such as Sisense to develop generative AI capabilities and enable customers to build generative AI models and applications, vendors should not ignore traditional AI, such as predictive modeling.
Enterprises still allocate half of their AI budgets to traditional AI, according to Ventana's research. Vendors, therefore, need to continue investing in enabling traditional AI development. In addition, they need to bring generative AI and traditional AI together so that each can make the other stronger.
"None of the BI vendors have really figured out how to bring these two worlds together," Menninger said. "It's an opportunity for software providers that can tackle that problem."
Eric Avidon is a senior news writer for TechTarget Editorial and a journalist with more than 25 years of experience. He covers analytics and data management.
