Image by: PhonlamaiPhoto, ©2017 Getty Images
When you think of machine-authored content, do you envision a room full of Arnold Schwarzenegger “authornators” sitting in front of computer screens? According to Gartner’s 2016 Magic Quadrant for Enterprise Content Management, “By 2018, 20% of all business content will be authored by machines.” Reaching that 20% won’t be hard. Machine-authored content is not a future technology. It’s a solution that is being used in many industries today. However, adoption of machine-authored content outside those initial solutions is where the opportunity for growth lies.
In some form, machine-authored content goes back to the 1960s. As banks started adopting computers, bank statements, detailing hundreds of individual transactions, were an obvious choice for machine authoring. Some early statements even included the use of analytics, leveraging tailored special offers to the account holder. These early forms of machine-authored content were very structured with rows and columns. Solutions like these still exist today.
Today, machine authoring starts with either compound document assembly or mail merge—not necessarily resulting in a physical document. The structure of these machine-authored documents is a document template, but they look very different from structured bank statements. These document templates identify the data needed either to complete key fields or to decide which specific sentences, paragraphs, or pages should be included in the document. The data passed to these documents can be used individually or as part of a complex logic formula. The same concepts can be used to define templates for document collections as well.
Automated Policy File Generation in InsuranceMachine-authored content has been used in insurance policy file management for the past 15 years. Many of these solutions are offered by traditional enterprise content management (ECM) or document management system (DMS) vendors. Typically, these solutions take an applicant's information to generate the cover policy form and select the various documents that need to be included as part of the policy file. These systems use this information to decide if insurance coverage will be provided and to generate costs associated with the policy. The individual forms are then machine authored to document this decision and provide signature lines for the application. These solutions also automatically select the appropriate policy documents to provide with the policy file. Often, these policy documents are also machine authored to include specific clauses that are appropriate to the individual policy holder. For example, a policy document on safe driver discounts may include specific state rules for verification based on where the policy was purchased.
Automated Drafting in Contract ManagementIn contract life cycle management (CLM), the top 20 vendors have some form of automated contract drafting. Most solutions use forms-driven contract templates where a request is completed manually with some pre-populated data from customer relationship management (CRM) solutions. Other solutions offer one-button requests from inside the CRM solution to automatically generate contracts.
The machine authoring, or drafting, of these contracts starts by selecting the appropriate contract template based on information the system received and completes the standard information needed, such as contacts, price, quantity, and dates. The machine authoring also ensures that appropriate clauses are included in these contracts. For example, a contract with a purchaser in New York may have different requirements than with a purchaser in New Mexico. Machine authoring would include the appropriate state clauses for each contract. Clauses may also vary based on organization size, products, or individual relationships.
These solutions remove the need for corporate legal to be involved in the drafting of every contract. In some cases, these systems include pre-approved alternate clauses that allow for quick re-drafting of contracts. This reduces the amount of time corporate legal may need to be involved during negotiations.
This machine-authored contract also supports the management of those contracts as well. Certain clauses may have obligations attached to them, like volume discounts. When these clauses are drafted, the author will indicate that an obligation should be created for monitoring.
Just the Beginning of Machine AuthoringThe examples in policy file generation and contracts are also being used in areas like mortgage applications and human resources. Machine authoring is also being used in large challenges, like industry equipment manuals, and common ones, like delinquency notices from doctor offices. Machine authoring is not coming; it’s here.
The role of the human author does not go away with machine authoring. As machine authoring solutions move from dynamically authoring paragraphs to sentences, new document authoring concepts will need to be developed. Authors will need to develop new skills to create dynamic sentences with proper grammar while maintaining the flexibility to include external data. The lack of this skill can be witnessed in much of the poorly generated junk mail we see today.
In the future, real content analytics, along with computational linguistics, will fully automate the generation of sentences and paragraphs. Content analytics will be used to assess meaning, while computational linguistics will bring the ability to generate properly formatted sentences. Even then, the human element will not be eliminated from authoring. After all, documents are a human form of communication, not a computer one.
Marko Sillanpaa is co-founder of the blog Big Men On Content and the founder of BMO Consulting. He has been working in ECM for over 18 years for vendors like Documentum, EMC, Hyland, and SDL Trados and systems integrators like CSC and Accenture. Follow him on Twitter @MSillanpaaBMOC.