Your content efficiency needs attention – Generative AI can help

Generative AI allows content teams to do more than they’ve been able to until now. It goes beyond conventional software solutions because of its flexibility, its comparative ease of implementation, and its ability to address a broader range of tasks.

An important concept in generative AI technology is the role of attention. A few years ago, Google researchers published a pioneering paper on generative AI entitled “Attention is all you need.” Generative AI bots direct attention toward content that can be transformed into new content — be it information, messages, or instructions.

Bots can focus attention on things your staff isn’t able to, so your staff can direct its attention to where it can be most useful. For people, attention is the scarcest resource. It’s better to let bots handle attention-intensive work.

Generative AI has the potential to improve the efficiency of a vast range of content operations. It can do more than create new content quickly. It can reshape content operations. 

Generative AI can improve the efficiency of all kinds of content. For example, it is already having an impact on news content. In a recent global survey of news staff by the World Association of News Publishers, “61 percent rated workflow / efficiency as an area where GenAI can help the most. In fact, 43 percent of newsrooms already use the tools for this purpose.”

The need to improve efficiency is everywhere. Much content work is manually intensive. Content processes may be poorly defined, resulting in ad hoc work. Or they can be disconnected, where decisions are made in isolation.  

By using generative AI in conjunction with a headless content management platform, content teams can re-engineer their content processes to attain greater efficiency.

Before teams can transform their current practices, they must first understand their dynamics. Many inefficiencies lay hidden because they have not been examined closely. Habits guide work processes. Teams perform tasks the way they were set up many years ago or the way they have always been done.

Generative AI offers the chance to break away from legacy content practices and rethink how content processes are done.

Phase 1: Assess the efficiency of your content operations stack

First, teams need to discover the sources of inefficiency in their content operations.

Content operations involve several layers, each of which specifies different levels of detail: 

• Content operations consist of multiple content processes — how organizations coordinate people with systems to change content — such as the overall process for creating new content, approving content, and getting it ready to publish.
• Each process will have one or more workflows, which define how to proceed given various beginning and ending states.
• Each workflow is defined by a set of discrete tasks, sometimes called workflow stages. 
• Each task will have one or more steps associated with completing the task.

When operations are viewed as layers in a stack, teams can probe the presence of inefficiencies at various levels of detail. By using generative AI for content tasks, teams are no longer limited to existing team resources.

Where within your processes do things get stuck or take too long? Think about the kinds of activities and sorts of people associated with activities that don’t run as smoothly or as quickly as you’d like.

Ideally, your organization has all its content processes and workflows fully documented, revealing how content moves through various scenarios. But in practice, many organizations will have at least some processes that aren’t formally defined or documented. 

If your organization doesn’t have everything documented, a useful way to get a handle on the characteristics driving inefficiency is to think about activities in terms of roles and responsibilities. All content tasks can be mapped to a RACI framework, a commonly used project management tool.

Think about the people and job roles involved with content, and ask who is:

• Responsible: who does the work?
• Accountable: who makes sure the work gets done correctly?
• Consulted: who needs to weigh in on the work?
• Informed: who needs to know the work is finished or know the outcome of the work?

The RACI framework can help teams assess content activities in two ways:

1. The RACI framework defines ownership of tasks and thus identifies the stakeholders (individuals or job roles) who are involved with activities.
2. Once the stakeholders are clearly identified, it becomes easier to identify the full range of tasks associated with content processes at all stages of the content lifecycle.

We now have a simple way to focus on how people and tasks are involved in current processes. The next step is to consider how these factors may be slowing down work.

Identify areas of inefficiency. Look at bottlenecks, which can arise from:

• People delays, often involving those with specialist expertise who aren’t available when they are needed or who are overloaded with work
• Detail misalignment, when tasks errors cause delays or rework
• Manually-intensive work that takes too long for everyone because it is time and labor-intensive

Ask:

• What tasks take the longest to complete?
• Which tasks involve the most interaction between various stakeholders?
• What tasks involve the most rework due to mistakes or oversights?

Even if your discovery phase yields mostly anecdotal information, it will still be helpful for suggesting how common kinds of situations could be improved using generative AI.

Phase 2: Identify potential opportunities for Generative AI

By developing an inventory of activities and their characteristics, we can start to think about how generative AI can help.

Starting with the RACI framework for describing current processes, we can extend it to imagine how generative AI could assume different RACI roles.

RACI dimension: Who is…	Common content tasks: What needs doing?	Generative AI opportunities: What might change?
Responsible for content	Creating content	How the content, metadata, and supporting code are created
Accountable for content	Checking content	How the content is checked, quality assurance
Consulted about content	Providing input on content	Applying best practices to the direction and requests relating to content through the design of prompts
Informed about content	Reporting about content statuses	The provision of feedback and notifications to staff and stakeholders

Generative AI can change workflows in three ways:

1. Streamline tasks by removing steps
2. Compress tasks by making them shorter in duration to complete
3. Eliminate tasks that currently exist as manual processes

What steps are currently not necessary when using generative AI?

Streamlining is achieved when fewer steps are needed. These iterations could be the number of steps an individual needs to do or else the number of interactions between individuals relating to a task. For example, if Generative AI can do a check in advance of sending a content item for review, it may reduce the number of interactions between the reviewer and the author.

Even when a task still requires the same number of steps, generative AI can accelerate the time required for completing the work. Manual steps are often laborious — sucking up too much attention — and time-consuming. Generative AI can do many steps in task nearly instantly, significantly reducing the total time needed.

When implementing generative AI, some tasks may not be necessary anymore — at least, not necessary for people to be involved with. Complete automation of tasks can increase speed and quality. For example, if formulaic content can be entirely written by a bot, then it will no longer be necessary to assign the task of writing that content to a specific individual. 

Bots can help with many procedural and mundane tasks within workflows and allow people to focus on tasks that require more judgment. Bots can be used to generate taxonomy tags or metadata descriptions for content, freeing staff to work on other priorities.

Changing the anatomy of content work

Generative AI can re-engineer how content processes work by changing what happens in workflows. It can reconfigure a range of tasks associated with developing, maintaining, and publishing content.

Generative AI offers three benefits for improving the efficiency of content processes. 

First, it can accelerate complex tasks by taking over the steps that are time-consuming. One of the most time-consuming aspects of content work is when various experts need to be consulted. Good prompt engineering captures the subject matter expertise and technical knowledge of various staff to enable that expertise can be reused to support future related work.

Second, it can change who needs to be involved by redistributing roles. Some tasks can be delegated to bots. Other tasks will still need people involved, but the use of bots means that fewer people need to be involved, and the task becomes simpler. In some cases, bots can remove the need for direct IT support for non-technical roles to complete the task.

Third, generative AI supports the automation of tasks by removing manual work. Automation is especially beneficial when staff currently need to manually process batches of similar items.

Phase 3: Set priorities for implementing generative AI

Teams have many possible ways to use generative AI in content operations. How should they decide what to focus on? Where should they direct their attention?

Content teams should balance coverage with feasibility. The goal should not be to introduce generative AI to every workflow or task — that would be impossible to accomplish in the near term. At the same time, implementing generative AI for only a selected class of tasks, such as revising wording, won’t deliver a large impact on content operations overall.

Many commercially available AI tools focus on targeted tasks that are narrow in their scope. These tools address the tactical level of content work by making individual tasks easier. But if implemented in a piecemeal manner, they won’t transform content operations.

When generative AI is applied to many tasks across a range of workflows, its impact will be appreciable. Since not every opportunity can be pursued immediately, teams should develop a vision of the scope they want to address using generative AI.

Treat Generative AI as a strategic capability that provides the greatest value when it’s implemented across content operations. Generative AI can process and transform text, images, and code relating to all kinds of topics and domains. The opportunities for greater efficiency using generative AI are ubiquitous.

The aim of efficiency is not just saving time and money, important as those outcomes are. Efficiency opens more opportunities for your organization. It allows your organization to do more.

Generative AI presents a chance to shape content operations proactively instead of focusing solely on fixing problems. It offers a way to reestablish content priorities. 

Your priorities for generative AI should reflect your content priorities. The content that’s most important to your organization deserves the benefits of generative AI first.

Prioritize generative AI opportunities according to their expected impact. Weigh opportunities against two factors:

1. The impact on resources and the ability to free up resources such as people and time
2. The impact on critical content

Use generative AI to address your blockers, the factors that constrain existing resources. Complex tasks are fragile. A small thing can go wrong that derails the process. A complex task is one that requires:

• Multiple people are involved
• A highly specialized skillset, limiting who can do the task
• A high portion of time locating or verifying information

Complex tasks are high-value opportunities for generative AI.

Prioritize the use of generative AI for critical content: content with outsized impact. Critical content often has one or two attributes that are essential to its success. For example, maybe the content needs to convey an update right away, or else the content loses its relevance. Or perhaps the content will only resonate with key audience segments if it is tailored to their specific situations — general discussions aren’t relatable to each segment, who see their needs as unique to them.

Ask what critical content is under-performing, and how generative AI can unblock it.  

Once you’ve identified some larger goals, think about specific content jobs that can be improved, such as:

• Better notifications to stakeholders to keep them informed
• Better feedback to authors to guide the development of content properly
• Better first drafts that need less editing
• Better awareness of existing content so it can be used again
• More precise content creation that incorporates specific up-to-date details
• Wider scope for tailored content

Develop a long-term vision. Adopting generative AI is not a fixed-term project. It needs to be an ongoing program. New possibilities and capabilities will be emerging all the time.

As a practical matter, not all worthwhile opportunities can be addressed at once. Rather than follow a set roadmap, it may be better to develop a long-term vision for using generative AI. The vision can capture the future state of how you want content processes to operate without getting too specific about which tasks to change or how to implement the changes — details that could shift by the time you can address them.

The below table shows some examples of “before” and “after” states. 

Before Generative AI (example)	After Generative AI (sample vision)
Stakeholders are unaware of the status and outcome of new content because they get vague communications or no communications at all.	Stakeholders are better informed about the status and results of new content through personalized, factually specific communications that explain issues that are relevant to their responsibilities.
The effort required to create tailored content limits our capacity to create content variations.	The ability to create many variations is now possible, enabling our organization to produce a wider variety of content tailored to different audience segments.

Making generative AI successful involves many details (we’ll explore some of these details in future posts). But be sure not to be lost on these details. Maintain your attention on the larger vision.