The Company

Krista Software was founded in 2016 and is headquartered in Dallas, Texas, with additional facilities in Pune, India. The company is led by John Michelsen, formerly founder of iTKO Inc., which was acquired by CA Technologies in 2011. Krista Software has raised around $20 million to date, most recently a $15 million Series A in February 2022 from new investors Grotech Ventures and iGrafx, joining returning seed round investor Rally Ventures. We estimate current revenues at around $10 million, with approximately 80 employees predominantly in the US and India.

The Technology

Krista’s eponymous platform is designed around the notion of efficiently connecting employees to information that enables them to make decisions and complete complex operations. The company believes that organizational adaptability suffers at the point where people, a diversity of systems, and complicated decisions converge, and that traditional automation approaches and technologies do not offer the best way to resolve this tension.

Instead, Krista’s “No code, AI-led Intelligent Automation” contains and combines familiar elements such as robotic process automation (RPA), low-code, and conversational AI to solve challenges of this type. In theory, at least, Krista provides simple user interfaces to otherwise complex automation designed to be built and managed by business users rather than developers.

At first glance, this suggests a “citizen developer” approach to automation. Still, a deeper dive reveals a more distinct, if not unique, angle to explore. Instead of empowering “citizens” to build apps, Krista provides an interface for workers to engage more holistically, directly, and as often as necessary with automation tasks and processes to structure and guide work activities rather than to build standalone process-centric applications. But what sets Krista apart from the get-go is its reliance on chatbots and conversational experiences; users can type in commands using natural language such as “escalate any inquiries from Deep Analysis to a supervisor.”

From a structural standpoint, Krista conceptually consists of three layers: infrastructure, the automation studio, and the end-user experience.

Infrastructure. At the base of the stack is an infrastructure layer consisting of connectors to legacy systems, RPA bots, AI and machine learning (ML) models, etc., along with, intriguingly, a blockchain element that can record and make immutable process transactions. This latter element is notable as, though there has been much discussion in the automation sector about the potential use of blockchain for audit purposes, there has been little to see to date. What we have here is Krista’s integration platform as a service (iPaaS) platform, which manages third-party integrations through a no-code interface. These ML models can be similarly called up as an Ask Krista option (including its own proprietary Cognitive Issue Resolution and Document Understanding models), as well as potential connections to existing automation platforms and organizational systems of record.

Automation studio. In the middle of the stack is the automation studio, where business users gain access to the system (see Figure 1). In the studio layer, described by Krista as “nothing like code,” users can express their desired outcomes and leverage the underlying AI and ML models along with accessing a catalog of predefined tasks, automation, and operations. These can come in many different forms; for instance, you can call upon information from third-party systems. Or you may call up decision-making information from Krista’s onboard AI (“Ask Krista”), defer to human beings (individuals or role-based groups) for a decision, or post a notification of an action’s completion.

For example, an interaction between an employee and Krista’s conversational AI might begin with a query about an internal procedure. The AI – interpreting the request through natural language interpretation – may be able to surface an immediate resolution by returning a link to a system or documentation that closes the inquiry. However, that query could result in required secondary questions, with branched-decision automation managed in Krista to handle the process. So, in short, automations can drive simple, repeatable tasks and more complex “if-then-else”-style decision trees, all built and maintained within the no-code automation studio interface. The interface can use pre-built actions and routing nodes, so that users can model a process as if it were a human conversation. This enables those who devise and are experts in the procedures to build and validate the automation that will underpin operations.

End-user experience. At the top of the stack is the end-user experience layer, essentially a set of APIs to integrate with such applications as Teams, Slack, SMS, or regular business applications.

On the one hand, Krista is a tricky automation platform to get your head around as it differs so markedly from process-modeling-centric BPM and single-task-centric RPA systems. It’s difficult to categorize, but that’s no bad thing, as Krista is straightforward to use, and its use of natural conversational language takes it beyond most no-code and traditional approaches to automation. It’s not an approach that would suit everyone or every occasion. Still, the company says that current platform users have found it particularly suited to use cases in security operations, contact centers, and restaurant operations. These are situations where rapid access to the proper procedure or customer resolution is welcome with the oversight of configurable AI-derived protective processes (such as anomaly detection).