Early-stage tech companies comb through data, software code

Boosting data quality in the industrial sector

Boston-based Aperio targets quality issues in time series data -- readings collected from industrial equipment over particular time intervals. Sensors and transmitters generate streaming data from machinery housed in power utilities, oil and gas companies and renewable energy firms, for example.

"The problem is there are so many point sources and data sources that individuals can't manage it all," said Jane Arnold, customer success lead at Aperio.

Organizations have been getting by with manual data-checking and processes for dealing with individual sensors. In Aperio's view, however, that approach becomes untenable for industrial enterprises that plan to pursue AI and advanced analytics at scale.

"When companies are putting tens of millions of sensor signals into these AI models, it becomes impossible," Aperio CEO Jonas Hellgren said.

He said the inability to check for issues in massive amounts of time series data through human inspection hinders the widespread adoption of AI in the industrial sector.

"'That is the Achilles heel of the industry right now," Hellgren said. "You need an unsupervised, autonomous solution for this problem."

To that end, Aperio's DataWise software provides machine learning models built to look for specific data issues. The models connect with a customer's data streams and inspect the data to find anomalies, Hellgren said. Those anomalies could include missing data, noisy data or miscalibrated sensors.

Feeback from CIOs and other IT leaders has helped Aperio refine its product. One request has been to integrate Aperio and its data quality insights into other frequently used tools. Arnold noted the example of Seeq's industrial analytics software for time series data.

"Many industrials are using Seeq for self-service analytics and advanced analytics," she said. "The biggest pull has been to integrate our events into Seeq analytics tools."

Automating go-to-market operations

Another Innovation Showcase finalist, iCustomer, aims to make enterprise customer data more actionable and ready for AI.

Customers generate an abundance of signals: purchasing history, web browsing behavior, demographic information and sentiment on their purchases and experiences. Those signals are recorded in a multitude of data stores, including an enterprise's internal customer systems and external sources such as social media platforms.

The problem for organizations is incorporating all that data into day-to-day decision-making, said Abhi Yadav, CEO at iCustomer in Cambridge, Mass.

"You have this customer data," he noted. "Well, how would you activate it?"

The iCustomer approach continues Yadav's earlier startup experience in the customer data platform (CDP) market. CDP products have made progress in collecting customer data from various sources. But Yadav said a gap persisted between static customer databases and the need to use data effectively in go-to-market operations.

"There was a CDP era, and now we call it a 'knowledge era,'" he said.

In Yadav's view, agentic AI characterizes this new era. His company's product uses AI agents to tackle tasks such as data unification, customer segmentation and marketing campaign analysis. In addition, the company provides an ontology, which describes relationships between people, companies and events such as customer support interactions. AI agents act upon this domain-knowledge ontology.

Yadav said iCustomer’s interactions with IT leaders have helped shape the company's positioning. It sells through integration partners such as data platform vendors Snowflake and Databricks. The idea is to preserve and enhance what the customer already uses rather than introduce a new product category, Yadav said. "One customer said, 'Look, the CIO has told us we are married to this data engine right now for the next five years. We have invested millions of dollars. We cannot throw this out the door.'"

Analyzing large, complex codebases

Silverthread's CodeMRI product line analyzes software code across industries such as aerospace and defense, healthcare and financial services. A key thrust is helping with software modernization, noted Daniel Sturtevant, founder of the company, which is also based in Cambridge.

"A lot of these codebases are incredibly complex," he said. "You can have hundreds of people, or thousands, contributing to their development over decades in some cases."

Sturtevant said massive codebases can become fragile, crystalline structures, making them difficult to modify. "Every time you poke it here, it breaks over there and you have no idea how or why."

Silverthread's tools and methodologies, based on MIT and Harvard research, make such monolithic systems modular and, thus, easier to work with and modify, according to the company.

The CIO perspective enables us to look at codebases from a more global perspective across an organization, rather than if you were looking from the bottom up on individual teams.

Silverthread's offerings also have an economic dimension. Sturtevant said the company's techniques can measure the health of a codebase, correlating the "size of the mess" with how badly it affects productivity, software bugs, security issues and time-to-delivery metrics.

A code analysis might find that an organization's software is in great shape or has challenges that would be economically reasonable and technically feasible to address, Sturtevant said. Alternatively, the analysis might uncover software that is too far gone for repair.

Measuring the health of software code helps organizations make -- and justify – strategic decisions, Sturtevant said.

In one case, a state government agency with 200 codebases was struggling to maintain its software, he noted. Silverthread provided its software and a consulting engagement and found that the agency's 15 million lines of code could be reduced to 3 million lines. After Sturtevant briefed the agency's CIO, the agency embarked on a consolidation plan to shrink the codebase to a size its workforce could manage.

From Silverthread's perspective, CIOs offer a valuable viewpoint that expands upon the observations of an organization's developer groups.

"The CIO perspective enables us to look at codebases from a more global perspective across an organization, rather than if you were looking from the bottom up on individual teams," Silverthread CEO Karen Chalfant said.

John Moore is a writer for Informa TechTarget covering the CIO role, economic trends and the IT services industry.