Workato partners with OpenAI to ease business automation

Join top executives in San Francisco on July 11-12, to hear how leaders are integrating and optimizing AI investments for success. Learn More

“Built using OpenAI models, the Copilot is like a Workato-expert coworker who generates workflow recipes and data connectors through a natural conversation. It has been trained on millions of data points from Workato’s public community,” Gautham Viswanathan, founder and head of products and engineering at Workato, told VentureBeat. “We believe Workato Copilot will further lower the barrier of who can build within an organization.”

Viswanathan added that the Copilot will assist users by providing onboarding support, learning new capabilities, discovering what to build next, offering recommendations, and providing instant troubleshooting help.

>>Don’t miss our special issue: Building the foundation for customer data quality.<<

Workato’s enterprise automation tool already incorporates RecipeIQ, its own AI/ML models, which provide data mapping, logic and next-step recommendations. By incorporating OpenAI’s models, Workato aims to further streamline automation and integrations development, making it easier for businesses to adopt its technology.

The company also said this collaboration ensures robust security and governance capabilities, enabling confident collaboration between IT and business teams and driving efficient operations at scale.

Streamlining enterprise automation through OpenAI models

According to Viswanathan, integrating OpenAI’s models into the Workato platform involved considering numerous use cases requested by its end users from various departments and industries. 

These use cases include functions such as generating highly personalized emails/sequences, summarizing meetings/recordings, and creating virtual assistants.

Since customers are currently building automations on the Workato platform, the company selected LLMs by evaluating these automations and envisioning how they can be enhanced through generative AI. The team then explored OpenAI models to determine which ones best suit each use case.

“This led us to select several LLMs and then train them with our proprietary models to best serve those specific use cases,” Viswanathan told VentureBeat. “We have seamlessly incorporated these models into our platform so that our customers can experience this as they build their automations, integrations, APIs, or application connectors.”

Workato introduced RecipeIQ in 2018, utilizing proprietary ML techniques to offer users recommendations for their workflow’s next steps. The company said that the Copilot will expand upon this feature, enabling it to construct complete recipes through conversational interactions with the builder.

Viswanathan said that the WorkbotGPT capability will facilitate real-time automation in business workflows, eliminating the need for pre-built components.

“WorkbotGPT is conversational automation for Slack and Teams. You can give it natural language prompts, and it will generate the summary of action items for you by looking up transcripts of recordings in Zoom, your email, and CRM — all in real time,” he said.

Ensuring secure automation development 

Workato said its platform incorporates a robust governance framework, facilitating the management of federated workspaces for different lines of business through AutomationHQ. 

The company also gives its customers full control over their assets, data and logs. The platform implements robust roles-based access controls and provides fine-grained permissions, allowing customers to determine who is authorized to use AI services. 

Customers can also mask sensitive data, audit all user activity changes, stream logs for centralized monitoring, and customize the storage duration of logs.

“For our international and multinational customers, we have multi-region data center support for customers that need to meet strict data residency and sovereignty requirements. Our Copilots adhere to the strictest data privacy standards and do not use customer data from these interactions to train any model,” explained Viswanathan. “These capabilities are built on top of a strong foundation of security featuring multi-layer encryption, hourly key rotation, EKM/BYOK, and zero-trust policies. “

What’s next for Workato? 

Viswanathan revealed that the company is presently training its models using metadata from user automations, integrations and internal APIs. The company aims to develop other powerful tools similar to Copilot and WorkbotGPT through this training. 

He believes that as enterprises increasingly embrace the power of AI, their trust in sharing data with external LLMs will grow.

“That will open a set of exciting possibilities — some we can think of, some will remain unknown until we fully understand the breadth and depth of available data,” he said. “We aim to solve that challenge by bringing AI, automation and integration to a single platform and creating new products and solutions that our customers can use to harness the power of these technologies.”