Domo to add prebuilt models, chat capabilities to AI suite

New AI capabilities

Domo first unveiled Domo AI in August 2023.

At the time, the suite included an integration with OpenAI’s ChatGPT, the general availability of Domo AI’s Service Layer so that developers can develop their own generative AI applications, natural language processing (NLP) capabilities, a set of software development kits and AI Model Management.

Now Domo plans to add AI Chat and Universal Models.

AI Chat is a feature that enables users to converse with their data, asking questions and receiving answers in natural language. In addition, the tool provides transparency, showing users the steps that were taken to come up with a response, such as the code that was generated and translated into natural language as well as which models were used.

That transparency was important to Domo when developing AI Chat, according to Ben Schein, the vendor’s senior vice president of product.

“AI is what is happening to our world, but the way in which you do it matters,” he said. “Some of the things you’ll see with how we approach AI have to do with focusing on transparency, showing our work and not being a black box.”

Some users might not care, Schein continued. But to others, knowing how a response was generated might be important.

Perhaps the biggest benefit of AI Chat is its potential to aid data exploration and analysis, according to Petrie.

Because AI Chat is able to offer suggestions as users converse with their data, it helps them find data they might not have otherwise realized could inform their decisions. It can also simplify the search for relevant data.

“[AI Chat] fits with recent findings by BARC and Eckerson Group, which show that adopters want to enrich their analytics outputs with GenAI and not simply boost productivity by automating repetitive tasks,” Petrie said.

In addition, AI Chat’s transparency stands to benefit users, he continued.

“There’s no question this is a priority for GenAI adopters,” Petrie said, noting that BARC and Eckerson Group have found in surveys that nearly or all respondents rank compliance concerns as a significant risk when generative AI is applied to business intelligence.

Mike Leone, an analyst at TechTarget’s Enterprise Strategy Group, similarly noted that Domo’s focus on transparency is significant. In particular, it fosters trust, he said.

“It’s not just Domo saying, ‘We released a new chatbot,'” Leone said. “The details matter immensely. Their focus on transparency is a key differentiator. Users need to trust these systems. By delivering information to users, that highlights the exact steps taken to answer a question and will go a long way in building that trust.”

Universal Models, meanwhile, is a suite of prebuilt AI models aimed at reducing reliance on data scientists.

Schein noted that many organizations don’t employ data scientists. In addition, among those that do, many don’t have large data science teams that can take on scores of different projects.

Universal Models, therefore, allows any organization to apply AI to a set of use cases. Among them are scenario planning, sentiment analysis, outlier detection and personal identifiable information (PII) detection.

Universal Models for forecasting — scenario planning — is scheduled for general availability in April through Domo’s Services Layer. Universal Models for sentiment analysis, PII and anomaly detection are scheduled for general availability shortly thereafter.

Leone noted that generative AI, including NLP, has been a dominant trend over the past year. Somewhat lost in the focus on generative AI has been traditional AI. Universal Models is therefore important for Domo users because it addresses traditional uses of AI.

“We’ve seen such an emphasis on GenAI capabilities that it feels as though the industry over-rotated,” Leone said. “Universal Models is about enabling folks to satisfy the use cases that are at the top of the list.”

Petrie, meanwhile, added that the significance of Universal Models lies not only in targeting traditional uses of AI but enables non-experts to build models.

“Universal Models mark a step toward democratizing data consumption by enabling more business-oriented managers to apply advanced analytics techniques to their datasets,” Petrie said.

Regarding why Domo developed AI Chat, Schein said customers were requesting more NLP capabilities. He noted that Domo offered some NLP capabilities through integrations with third-party vendors but did not enable full conversational interactions.

Customers, meanwhile, had been hearing about such capabilities as other vendors introduced generative AI tools.

“Everyone was asking about [conversational tools], so it’s checking a box,” Schein said. “But beyond checking it, we wanted to do it in a way that adds value.”

The impetus for Universal Models, meanwhile, came more from Domo’s own observations, Schein continued. In particular, the vendor saw how customers that lacked data science expertise struggled to build AI models so is now attempting to provide them with a foundation.

Additional features

In addition to the introduction of AI Chat and Universal Models, Domo unveiled the general availability of AI Model Management and ResponsibleGPT in Domo.AI.

AI Model Management aims to simplify managing models hosted outside of the Domo environment. That includes customer-developed models hosted in OpenAI, Databricks, Amazon Bedrock and Hugging Face, among other generative AI platforms.

Once deployed to AI Model Management, users can access the models and integrate them with their own existing machine learning infrastructure. In addition, AI Model Management enables customers to train Domo-hosted models with Jupyter Workspaces — web-based environments based on an integration between Domo and the open source Project Jupyter — and AutoML.

Domo’s AI Chat capabilities seem to target … users well, not just by responding to prompts but also by suggesting questions to ask and topics to research.

ResponsibleGPT, meanwhile, is an application aimed at ensuring that customers’ data does not get shared to public large language models when organizations integrate with LLMs such as ChatGPT and Google Gemini to derive insights from their data.

The application was built with Domo’s text generation AI Service so that the connection between customers’ data in Domo and LLMs is made through an API. The API restricts the flow of data, which not only prevents customers’ data from getting stored in LLMs but also enables data administrators to audit and review interactions between their organization’s data and LLMs.

Beyond the evolution of Domo.AI, Domo made App Builder generally available and updated Cloud Amplifier.

Cloud Amplifier, meanwhile, is a multi-cloud environment that provides native integrations between Domo and cloud data storage platforms such as Amazon Redshift, Databricks, Dremio, Google BigQuery and Snowflake.

Using the integrations, Cloud Amplifier enables customers to take data from one storage platform and enrich it with data from another storage platform.

The updated version includes the general availability of Magic ETL, a drag-and-drop tool that enables users to query and transform data where it resides.

App Studio is a low-code application development framework that enables customers to build customized applications, including customers who use Domo as the engine for their own proprietary applications that they sell to third-party customers.

App Studio features include the following:

• A new theme engine that gives users control over the look and feel of the applications they develop so the applications match their own branding styles.
• A pro-code integrated development environment in addition to the low-code development framework that enables trained developers to exert more control over their development experience.
• Domo AI capabilities that allow users to generate relevant content with a quick prompt and suggest additional content to further inform applications.
• Inline editing and full create, read, update, delete capabilities when developing applications.

“The emphasis on personalization is important,” Leone said. “Domo is creating an environment that can really cater to personal preference while empowering folks based on their level of comfort, expertise and goals.”

The idea behind App Studio was to provide any Domo user with the tools to build an application, according to Schein.

The low-code capabilities cater to less refined users who don’t have the skills to develop applications from scratch. In addition, they can be used by hardcore developers as starting points. Meanwhile, the pro-code capabilities provide those hardcore developers with a development environment for creating sophisticated applications.

“The genesis of App Studio was to build an environment where anyone can build a customized application,” Schein said.

Petrie, meanwhile, noted that it will be interesting to see exactly how Domo integrates Domo.AI with App Studio.

“Enterprises continue to blend analytical functions into applications and workflows, and App Studio assists this trend,” he said. “I’m interested to see App Studio now uses Domo.AI to assist code development. Software development has emerged as a popular, if not the most popular, use case for GenAI.”

Future plans

Looking ahead, Domo’s roadmap centers around four main ideas, according to Schein.

The vendor wants to develop a more composable ecosystem, which includes adding new partnerships and integrations so customers can pick and choose which tools to connect to Domo. In addition, Domo plans to continue adding AI capabilities and tools that enable customers to develop AI applications. Improved navigation of Domo is another focal point for the vendor. Finally, providing efficient data and AI governance is critical.

Regarding the addition of more capabilities related to AI, Domo integrates with OpenAI as a default, Schein noted. Customers can switch to the model of their choice. But beyond that, Domo is in the process of developing DomoGPT, its own LLM.

Unlike integrations with ChatGPT and other public LLMs that require user to import and export data, DomoGPT will enable customers to do all their generative AI development and analysis within Domo’s secure environment.

“We’re not trying to be OpenAI,” Schein said. “But … this will be an LLM only within the Domo ecosystem where our customers have already gone through long security reviews.”

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.