Job Redesign Around AI: Work Intelligence Tools Arrive
Let me posit an important question: if your CEO or CFO asks you to use AI to improve productivity, what would you do?
Would you use the “chainsaw” model of efficiency and lay off 10% of the employees? Or would you thoughtfully redesign jobs around AI?
This week I talked with dozens of companies about this topic, so I want to share what we’ve learned.
Why Companies Become Bureaucratic
Let’s start at the cause: how bureaucracy was created. It’s really pretty simple: as the organization grows, managers keep hiring new staff, often in support or administrative positions. Suddenly you wake up and your company is filled with “program managers,” “analysts,” and “project managers.”
We recently looked at the job architecture of a large tech company and almost a third of the jobs appear to be staff positions, analysts, or project managers. While I’m sure these folks are busy it’s clear to management many of these could be centralized, shared, automated, or eliminated.
We’re working with a large media company and they’ve been analyzing the work of their “media managers.” With almost 7,000 staff in this role, the myriad of tasks and activities these people perform is staggering. At the core they buy ad space for clients, but in reality they do creative design, account management, analytics, and must also keep up with AI.
I would call that job a “pivotal role” (one that drives outsized value), yet before today there was no standardization in place and the team now knows their new AI platform could transform the function.
How Do We Redesign Work: A Blueprint
Job design has been going on for many decades, and much of it starts with “Job Task Analysis.” Prior to AI (and tools like Reejig, Draup, and Gloat), we would do surveys to figure out what people are doing and then spot areas of overlap, inefficiency, or automation potential.
But in the age of AI, this may not be enough. Today, as AI automates everything from recruiting to needs analysis to content development, ideally we’d have a broader approach. And this is what I see companies struggling with.
Consider the following Blueprint, an approach that prevents you from using AI as a “solution looking for a problem.”
 |
First, you benchmark your company to see where inefficiencies exist. Tools like Reejig and Draup make this easy, and they give you a big look at engineering, sales, marketing, or individual roles to see where you’re overstaffed.
- The media company I mentioned above has seen its revenue and margins drop, so they’re focused on a client-facing role with a goal to improving growth.
- IBM spent the last decade using AI to automate many of the talent and performance processes in the company, and now has an agent that acts as an HR business partner. The AI decides pay bands to maintain pay equity, also giving managers in-depth data for performance reviews. The result is greater trust, reduction of bias, and an elevated design role for HR teams.
- Allianz has spent years optimizing its claims process. Understanding the critical role this job plays in profitability, they’ve now built a “digital twin” to automate and standardize much of this work.
As you can see, these may not be “bottoms up” but rather “tops down” projects. In many cases, when the CEO wants to do a layoff for example, this is the way to go.
Second, you now have to “decompose work” to figure out where AI can play. In some cases you look at a tool like SeekOut Spot or Paradox (state of the art recruitment agents) and you just “implement it” and rethink jobs. But that creates lots of fear and pushback, so it’s often better to work systemically.
Analyzing Tasks (or Activities)
Imagine you work in a hospital and you clean the floors: your “skills” went from “sweeping” to “operating the cleaning equipment.” If you’re a software engineer your skills are changing from “coding” to “using Github Copilot.” Marketing professionals are transforming from “creating campaigns” to “operating the AI creative platform.” And instructional designers are moving from “building courses” to “prompting AI and curating content.”
Once we know what these “tasks” or “activities” are, we can predict or decide how much automation to implement. In every case we’ve discussed this falls into four steps.
- First, is this team unproductive because they’re working on a product, sales process, or other initiative we may not want to do at all? I’ve worked in many low-productivity sales teams where the problem was a product nobody wanted to buy, not the sales process itself.
- Second, are the work tasks routine and easy to outsource? Could we centralize them? Are their easy-to-use platforms already in place?
- Third, if we find an AI tool to use, how hard will it be to build, optimize, and train? There may be some off-the-shelf offerings ready, but in some cases you may need IT support to build a system you need.
- And fourth, if we do outsource or automate these tasks, what new value-add will people need to learn? If a marketing professional is suddenly replaced by a CRM tool, for example, is he or she ready to become a strategist and add value on top?
Go back to the media company again. Your staff builds creative campaigns, buys advertising space, and manages e-commerce and activity metrics to continuously improve your client’s brand and sales performance. There are hundreds of “tasks” involved in this, including everything from client management to campaign management to various forms of creative work, managing events, doing SEO analysis, and much more.
As “agents” come along, your agency doesn’t want to fall behind because one of your competitors may suddenly under-bid you and out-perform you with your clients. So you’re going to be very interested in “what tasks and activities can we automate?”
Now the big question: do we have tools to help us break work down into tasks and figure out what to redesign? Yes, that’s what’s coming next.
New Work Intelligence Platforms From Reejig, Draup, Gloat And More
If you go back and think about the problem, this too is a “big data” opportunity. If I built a tool that scanned every job posting in the world and looked for “tasks” not just “skills,” and then matched these “tasks” against the job description, I would essentially have a massive “job task library” that updates itself in real-time. This is what Reejig has done.
You can use the Reejig platform to look at the jobs in your company and it will give you an accurate picture of “what tasks people are doing.” Microsoft and WPP are now doing this and they all told me the accuracy is amazing. In other words, while your company is not precisely identical to others, the real work people do in each business area is amazingly similar.
Draup, led by the founders of Talent Neuron, goes further. You can use Draup’s data and tools to benchmark your organization size, skills in place, and roles against peers. You can then look at individual roles or job families and see what “workloads” or tasks they do, with a set of tools to help you find opportunities for automation.
Job design is typically done by line managers. Throughout my career (sales, marketing, product management, etc.) I’ve always found that each company operationalizes things differently. So if you don’t do “job design” carefully each line leader will create roles that they believe are needed. The result is a sprawl of jobs and roles, often not taking advantage of automation tools already in place.
Gloat is doing something for these managers, giving them a system that decomposes work into repeatable activities based on Gloat’s task model. Imagine you are a marketing leader and you have to create a large event.
Gloat’s new Mosaic platform would literally help you do job design in real time, showing you the steps (or activities) commonly used to complete that process or project.
When I first saw Mosaic I was stunned. For all the years I’ve done sales, marketing, and research I’ve relied on my own experience to know what to do. Gloat could tell me every step to consider. And in case I’m not familiar with one of these steps, Gloat could find me a person with the skills needed to do it. And this idea is going to happen throughout our companies.
Consider this: while every business process in our companies feels similar (ie. “order to cash” or “leads to sale”) the details within these processes change all the time. Think about the process insurance companies use to process claims. Can you imagine how many validation steps, benchmarking, quality checks, and fraud detection processes they need when your house burns down? The folks at Travellers told me they have “simulated homes” and they actually destroy kitchens by fire to figure out what contractors, supplies, and costs they should cover in a fire.
Everything we do in business is made up of tasks and activities, and these “work steps” are getting automated at light speed. So these new “Work Intelligence” tools are going to be very useful and important in the world ahead.
In many ways tools like Reejig, Gloat, and Draup, and others are the new Work Intelligence tools we need, replacing much of the “job-task analysis” we’ve done in the past.
This is a new market, and it’s just beginning. More of these tools will come.
Final Thought: Think Business Redesign, Not Just Job Redesign
The “Rise of the Superworker” effort is both a business and job design project: we are integrating multiple job functions into data-driven agents.
One of our clients, for example, came to us recently and exclaimed “we think we have too many people and we want to double the size of our company without hiring anyone new.” (The “talent density” strategy.)
We looked at benchmarks (revenue per employee, etc.) and found that they may have 10-15% too many people. But before we dug into the job architecture, the CHRO mentioned “I think we have too many sales people because we’re selling to the wrong customers. Many of our small customers don’t renew.” No amount of job redesign can fix that!
In other words, the redesign of work is both a top-down exercise as well as a bottoms-up automation project. So remember the Blueprint and consider four things:
- Are there products, markets, customer segments which we should cut, fix, or rethink? Are we building the right products and serving the right markets?
- Are there skills and “misalignment” problems we can solve through training, shared services, or organizational integration?
- Are there repeatable, routine, low value tasks we can automate and streamline immediately? Can we quickly automate or streamline these using platforms already in place?
- As we move to automated agents, are there cross-functional opportunities to improve multiple roles at once?
If you think about recruiting, you could easily find a tool for building job descriptions. But if you consider the entire process, a multi-function agent could help with job analysis, job description, sourcing, assessment, and onboarding. (Paradox, Maki People, Eightfold are doing this today.)
Don’t Let AI Be A Solution Searching For A Problem
One final thought from my meetings. Some companies are so enamored with AI they feel like a solution looking for a problem.
New automation solutions will take time to implement, so go slowly and make sure you’re focused on high-return areas. That way you’ll get funding and IT support for this work.
 |
One final story.
A client is a large technology company with a global HR organization. They have excellent HR tech and productivity is already high. The “problem” they’re chasing is the ability to superpower employee growth as the business moves to an all-AI product model, and they want the HRBPs to lead this charge.
With this focus in mind, the HR team examined the inquiries, interactions, tasks, and activities these business partners do. Sure enough, through analysis with Reejig, they found as much as 40% of their time is bogged down with administration. Now, focused on the high level objective, the team is looking at automation (including Galileo) to automate these tasks.
Focus your job design efforts on a business objective. New tools like Gloat (job decomposition into projects, tasks, skills, and talent), Reejig (task analysis and organizational task benchmarking), and Draup (enterprise wide benchmarking, workload analysis, and technology platform benchmarking) help you speed the work.
Let’s not get enamored with AI for its own sake and stay pragmatic from the start. Today’s economic realities demand it.
We are here to help you sort this out, call us if you’d like some advice.
Additional Information
The Organization Design Superclass: Certificate Program
Busting Bureaucracy: Are Layoffs The Only Way To Go?
The Road To AI-Driven Productivity: Four Stages of Transformation
Galileo™ Professional, The Essential AI Assistant for Everything HR