The next automation frontier is unstructured data

The automation market is changing quickly. Yet, what is the context for all those automation discussions? Through which lens should we look at automation? Is it robotic process automation (RPA)? Is it an extension or evolution of RPA? Or do we have to look through a fundamentally different lens at all those discussions? HFS has repeatedly stated that automation is not your strategy. It is the necessary discipline to ensure your processes provide the data—at speed—to achieve your business outcomes. Against this background, it was refreshing to discuss these issues with a provider with a fundamentally different lens. Automation Hero puts artificial intelligence (AI) and unstructured data at the heart of its strategy with the potential helping enterprises finally make broad automation progress.

Focusing its automation on AI and unstructured data differentiates it from RPA

Automation Hero, headquartered in San Francisco, flips the automation discussions on its head. In its view, the intent is not to expand automation technologies such as RPA or runbooks with a plethora of adjacent technologies but to automate as many process steps with AI as possible, from data entry to reaching out to customers at the perfect time. It developed its Hero Platform_ as a new business operating system combining fast data, AI, and business process management capabilities to form an intelligent business automation platform optimized for workloads with an AI-centric mindset. Automation Hero’s approach is about ingesting data and using AI to enhance its value.

Much of the RPA-centric automation discussions focus on automating structured data, such as invoices, where AI modules help to identify known items, such as addresses or billing information. Unstructured data is information that either does not have a pre-defined data model or is not organized in a pre-defined manner. That includes obvious examples such as email. But again, the differentiation of Automation Hero does not lie so much in IDP use cases, but broader data capabilities. Real-time data monitoring and meta data scanning stand out in that regard. Therefore, it can be challenging to store that data in databases. Here, Automation Hero’s AI capabilities come to the fore.

Training Deep Learning models is at the heart of Automation Hero’s approach

Automation Hero’s approach focuses on a set of AI capabilities. Most are pre-trained models that can be either configured (no training) or advanced (train additional neural network layers with domain knowledge). Hero Platform_ tightly integrates with Google TensorFlow to train and execute a wide range of deep learning models. Existing models can be installed and used as a function in any data or business process workflow and take advantage of the enterprise readiness of the platform. The Hero Platform_ can train pre-defined (e.g., in Python) models, or its own AI design studio can create deep learning models. The platform contains three advanced deep learning-based engines:

• Context Aware OCR: Advanced OCR for structured, semi-structured, and unstructured documents with high accuracy.
• Recommendation Engine: Recommendation engine for cross and upsell, churn prediction, and best next-step suggestions.
• Classification and Intent-detection Engine: Identifies the intent of written communication, including emails, text messages, or online forms and posts. This engine can also classify documents or emails and route them or apply specific extraction strategies.
• Dark Data Extraction Engine: Can be trained to extract structured data from unstructured data. It can extract names, job titles, phone numbers, and address or product data from text conversations. It also can be extended to first convert (e.g., image data to text) and then extract meaningful structured data. The latest version of the extraction engine does not need to be trained but can be configured with natural language questions, which gives Automation Hero a strong differentiation.

IDP is a landing patch that is accessible by the broader market

Refreshingly, Automation Hero is largely staying clear of trying to bandy around monikers such as RPA and intelligent document processing (IDP) to gain visibility or traction. However, it uses IDP as a pragmatic way to achieve two things. First, it is a landing patch with clients to demonstrate its capabilities in a “low-level” use case. Second, once clients have trained their staff and captured value from their investment, upselling to much more complex scenarios becomes easier. All too often, AI approaches that go beyond just machine learning provide a barrier to entry through perception issues, a lack of talent, or insufficient training data.

Typical use cases are claims processing with strong traction within insurance. The other cluster of use cases is around supply chain optimization. For some clients, the platform gets trained to find POs even if they are not in an invoice but in other relevant documents. Furthermore, the platform can extract billing data from non-standardized orders, even for highly specific items. The ability of the platform to understand the context and receive that information in a knowledge graph is crucial. This ability allows for a non-linear flow of information that is conducive to the vast amount of unstructured data and the complexity of cloud-native operations. Automation Hero creates a knowledge graph by connecting a data source with historical data, for instance, ERP or CRM. Then, it internally builds a graph to improve OCR accuracy, classification, or other AI models.

The goal is an operating model for business processes

Automation Hero’s goal of evolving toward an operating model for business processes is ambitious. Yet, as Exhibit 1 outlines, it is a compelling example of the evolution of automation where process orchestration and automation enhance workflows, and process intelligence drives the execution of data and insights. Thus, Automation Hero’s competitors are not just IDP providers but the new cool kids on the block, like ServiceNow, Workato, Soroco, and Turbotic. The differentiation will come by demonstrating capturing value and progressing toward end-to-end automation. The battle lines are drawn here.

Exhibit 1: Automation Hero aims to evolve into an operating system for business processes

Source: HFS Research, 2022

The Bottom Line: The effective ingestion of unstructured data will drive more value in automation deployments.

Markets are won by sales and marketing, not capabilities. Thus, Automation Hero must invest in marketing to highlight its differentiated capabilities. Part of that is developing a partner model that focuses on sales and customer success in equal measure. Enterprise executives should evaluate the platform’s effectiveness while comparing it to IDP and much broader data ingestion. The value proposition is to provide actions on insights. The next battleground is unstructured data, not just the digitization of structured information.