Powering the cognitive digital thread with GenAI to accelerate innovation

Excerpt from the article bylined by Xavier Pornain published in « The Manufacturer » magazine.

Microsoft CEO Satya Nadella has called AI-powered search the ‘biggest thing for the company’ since the development of cloud 15 years ago’. However, how is this new technology making waves in the manufacturing sector? Xavier Pornain, SVP, Global Manufacturing at Sinequa.

One of the most common issues for businesses, particularly those working in large organisations, is locating and accessing their data. In 2022, a joint survey conducted by Sinequa and CMSWire found that a third of employees can’t find the information they’re looking for in their workspace, which includes heterogeneous file repositories, enterprise systems, etc.

With growing external pressures, such as decreasing time to market and threats of disruption to the global supply chain, leaders need their teams to be operating effectively and can no longer afford to implement inefficient processes and technologies. This is where innovations in AI are taking centre stage.

What are the common issues facing manufacturers?

For manufacturers, connecting users with relevant data is of particular concern. Most large manufacturing organisations have multiple Product Lifecycle Management (PLM) systems from leading vendors like Dassault Systèmes, PTC, or Siemens that categorise and organise their product engineering information. The search function in many of these applications is limited to full-text search with limited use of ontologies and semantics, so there is little opportunity to discover data that users are not explicitly searching for, but that may be related to their query.

When you add in the fact that a lot of the product information resides outside of PLM environments, for example, in collaborative apps, shared file systems, other enterprise systems such as the CRM or ERP, or other repositories, the task of locating relevant content becomes even harder. This means that users have additional hurdles they need to jump from one enterprise application to another that can be a source of frustration for employees and ultimately can have a negative impact on productivity.

The problem is not limited to frustrated employees but is also a hindrance to operational efficiency and time to market. Within the manufacturing industry, this can have a domino effect triggering:

• Longer Product Development Cycles: Delays in accessing critical data and information impede your team’s ability to progress with product development, leading to unreliable timelines.
• Quality Issues: You’re likely to experience an increase in the probability of quality issues if your team cannot access previous research and information. This could potentially translate into financial costs for your business.
• Increased Resolution Time: If the time spent searching for information increases, so does the time until you get the right answers to your questions. This can impact lead time, customer satisfaction, and product quality.

Furthermore, when combined, these hindrances and access challenges are likely to reflect poorly on the end product and your company’s reputation and could negatively impact customer satisfaction.

How is AI being deployed to make an impact?

To overcome these problems, manufacturing leaders are looking for innovations in artificial intelligence to bridge the gap. AI-powered search is a key component for CIOs, IT leaders, as well as VPs of Engineering, and VPs of Customer Success to implement a successful digital thread strategy. Digital threads permit the creation of a continuous flow of product information from design through manufacturing and support. With the technologies available today, a cognitive digital thread spans the entire product lifecycle within a company and extends outward to suppliers, customers, as well as products, and people in the field.

Over the last year, generative AI has become a front-runner in workplace technology. However, alongside its benefits, it has come with an array of issues. Data confidentiality, hallucinations, and a lack of visibility into the data sources are just some of the limitations businesses fear could impact their bottom line.

However, this is where AI-powered search, such as neural search, is leading the way. Neural search models are designed to use Natural Language Processing (NLP) to scan enterprise data and generate accurate results to questions. This can give users specific responses using their company data, reducing time to insight and empowering all employees to accomplish their jobs efficiently.

In addition to this, some leaders are making use of tools that combine neural search with the benefits of generative AI. These models feed the accurate results generated by neural search into a Large Language Model (LLM) to produce automated summaries of the findings. This presents engineering teams with fast, accurate answers on parts and products they are working with across all possible sources of information and, ultimately, streamlines workflows and enhances collaboration across departments. Due to the use of Retrieval Augmented Generation (RAG), this can be done without leaking any company confidential information outside the organisation while benefiting from the power of the LLM model.

In the same way, maintenance and support teams can also use the technology to find relevant answers within large technical documents as they test procedures. Instead of being tasked with reading all potential responses, scanning each individual document, and manually combining them into one report, these tools can help users with summaries based on the accurate and traceable data specific to their enterprise. This is of huge benefit to firms looking to make use of generative AI technologies but avoid the associated pitfalls regarding accuracy and hallucination.