BCG execs: AI across the company increased productivity, ‘employee joy’

With more than 100 offices and 32,000 employees, Boston Consulting Group leaders knew their decision to roll out several artificial intelligence (AI) platforms over the past two years meant not only deploying the technologies but training the firm’s workforce to use them.

While some organizations have chosen to train up a portion of their employees who might most benefit from using AI, BCG opted to train its entire workforce on using generative AI (genAI) platforms. It believed offering the tech to everyone would allow efficiency and productivity gains to grow organically.

Massive upskilling is essential to delivering broad efficiency gains from genAI. And a program that conveys how fundamental the technology is to employees’ effectiveness and longer-term career growth can be key to winning their hearts and minds.

The investments paid off. BCG execs say employees are more innovative, they’re finding their own efficiencies, and overall are more happier about how they work through the productivity gains of AI.

Foundational to BCG’s AI efforts was its enterprise GPT strategy. Beginning in October 2023, ChatGPT was rolled out internally to every employee. But the technology was kept in house, meaning all data remains under BCG’s control.

BCG allows its consultants to build their own GPTs for specific customer interactions, which has fostered an atmosphere of innovation. To date, more than 6,000 GPTs have been created by BCG’s employees to perform tasks such as summarizing documents and video meetings, and automatically generating email responses to clients.

BCG also developed Gene, a conversational AI chatbot that can interact with humans. Gene was originally used as a co-host for BCG’s “Imagine This” podcast, but it evolved into a client engagement and content creation tool. Gene’s memory banks are stocked with research and insights on genAI from some of BCG’s top analysts.

Alicia Pittman, BCG global people chair, and Scott Wilder, BCG managing director and partner, answered questions from Computerworld about the firm’s AI strategy, what it took to implement, and what the benefits have been. The interview has been lightly edited for clarity.

Boston Consulting Group

I understand that BCG rolled out AI tools such as enterprise GPT and Gene to assist your workforce in performing its tasks. Can you explain how those AI tools are being used at BCG?

Alicia Pittman: “Our philosophy around genAI at BCG is ‘early and often.’ We understood from those first days that genAI would be transformational in how we work. And we knew that to help our clients transform in this space, we needed to first transform ourselves. We rolled out best-in-class tools in the market — including ChatGPT Enterprise for all our staff — and also built our own, including a tool called Deckster to help generate slides and a natural language query search for our knowledge base.

“To date, our people have created over 6,000 custom GPTs for various tasks and client work to enhance productivity and simplify administrative duties. And since our global launch of Deckster in March 2024, it has helped us create or edit slides over 450,000 times. We want our people to be ahead of the curve so we can be at the cutting edge as the technology evolves and expands the art of the possible. Being at the forefront is a promise we make to our people and our clients — and we’ve invested accordingly.”

What is a “custom GPT” and what are examples of tasks they can tackle?

Scott Wilder: “A custom GPT is a powerful feature within ChatGPT that allows someone to build a reusable mini genAI application that users define themselves through natural language. This is done through a combination of custom instructions, files you can load (e.g., best examples, or data, etc.), and an ability to program interactions to address specific tasks. Our teams have developed and shared more than 1,000 custom GPTs on everything from how to improve ChatGPT prompts to scheduling meetings across time zones to generating more creative talk tracks for trainings.

“There are a little over 6,000 custom GPTs total. Within this pool of custom GPTs are about 5,000 private ones for individual BCGers and about 1,000 that are shared among BCG teams. Private GPTs can only be accessed by the person that made it (to solve for something specific or niche that only they were dealing with). Shared GPTs mean that the maker can share them with whoever they want, but only within BCG. They can select specific people across case teams or make them ‘public’ within BCG.”

When did BCG begin its internal rollout/deployment of AI? What was the purpose of the rollout?

Wilder: “We started evaluating genAI tools in late January 2023 and, in tandem, assessed the top activities to drive greater efficiency within, when genAI was first taking the world by storm. Our first major release under the rollout was ChatGPT Enterprise in October 2023. At the time, it was ChatGPT 3.5. We also custom-built tools like Deckster around that time to deploy in beta to help save time creating and editing slides.”

In what ways did ChatGPT and other AI tools increase performance and efficiency?

Pittman: “We are seeing a wide range of applications across our business and with our clients — from how we ramp up new team members on complex cases, to accelerating market research, to more rapidly customizing client collateral for higher-impact implementation. For instance, with one client we are equipping the sales force to roll out new methodologies. We might have typically provided one script for each of the 8 to 10 different client segments for the sales team to then customize. With genAI, we can provide 100, 200, or even 500 custom scripts at the customer level.

“Equally, genAI is upping our game in how we develop people. New managers at BCG are using avatars to practice giving high-quality feedback. GenAI also helps our people synthesize multiple types of feedback to sharpen their skills and focus areas. Our teams use a mix of off-the-shelf and custom-built tools.”

Wilder: “Putting all this together, genAI is really improving the quality of our insights as we have more time for more important tasks, improving the breadth and depth of what we can provide to our clients, and accelerating the learning curve, which will have quality benefits for years to come.”

How did BCG measure AI performance/efficiency metrics?

Pittman: “We have three primary KPIs [key performance indicators] when it comes to our genAI transformation: increase productivity, enhance the quality of our work and insights, and increase employee joy by reducing time spent on toil. We are seeing high impact across each of these, which is exciting for our people, for our clients, and therefore for our business. As we ramp up this year, we are also measuring usage and inputs carefully. In particular, we are measuring what features and habits spur stickiness in usage.”

How did BCG go about retraining its workforce to use its new tools?

Pittman: “We started by doing a 360 review to understand the capabilities, desired outcomes, and potential risks of the tools we were experimenting with to ensure the right guardrails were in place before fully rolling them out to our employees. Responsible and ethical use of AI is something we take very seriously, so that was an important early consideration. We have strict data control measures and set up various trainings. We keep those updated on a rolling basis as the technologies evolve.

“Approaching the rollout, we took our own advice and really invested in what we call ‘the 70.’ We have found that successful AI transformations typically dedicate 10% of their AI effort to algorithms, 20% to data and technological backbone, and 70% — the lion’s share — to business and people transformation.

“So, we have had to fundamentally rethink the ways that we work. That looks different for everyone based on the nature of someone’s day-today job and what they are looking to do. Some BCGers find the technology most useful for creative content generation, others for getting new team members up to speed quickly, and others for knowledge gathering. For example, one of our executive assistants (EA) built a tool to automate tasks for himself and other EAs. The underlying need dictates which of our tools they turn to.

“We deployed an ecosystem of training and upskilling that includes everything from our formal required curriculum for BCGers at all levels, to on-demand virtual trainings, to one-on-one reverse mentorship for our superusers to help them coach others. It’s been an intentional mix, recognizing that different people learn in different ways.

“Our work is so team-based, we knew that just working with individuals would not be enough. We have team coaches that help our teams find and develop applications in real time. We also believe that change happens at the rock face, so we have individual and team coaches available to help people on the spot in their day-to-day work.

“And we work to make it fun. Our workforce includes some of the brightest and most innovative talent out there. So, we let that shine. We ran a ‘GenAI Olympics’ to identify the best ideas bubbling up from our teams using these tools at the rock face. We are constantly learning and adapting as we go — in addition to building the big upskilling programs necessary to make the change stick, at all levels of the organization.”

Did you retrain your entire workforce, and how did training differ depending on worker roles?

Pittman: “We made a deliberate decision to make our tools accessible to all roles out of the gate. We then created customized communications and training programs for job types and cohorts — we want these tools to be useful, and similar job cohorts often have more similar needs.

“That said, we also encouraged significant peer-to-peer learning and sharing. We rapidly built a GenAI Enablement Network across various job types that champions local innovation and training. This network is made up of 1,200 BCG employees from around the world who have volunteered to be a part of activating genAI across BCG. They represent each of our offices across 50 countries, as well as each of our businesses and functions. Their activities include mentoring, hosting upskilling sessions, and sharing innovations. 

“The Enablement Network is a cadre of passionate genAI users who promote local and global genAI initiatives while driving adoption and transforming our ways of working across the firm. We are all on this journey together, and learning and expertise can come from all directions.”

What were the largest hurdles you experienced in rolling out generative AI at BCG? 

Pittman: “BCG has a strong culture of innovation and learning. So, our broader challenge has been to roll out AI and other associated tools fast enough! Broadly speaking, our people are eager to experiment and gain expertise, and that has certainly been the case for genAI as well.

“Two areas that we have put extra focus on within our rollout are responsible AI and ensuring that the adoption of this technology is inclusive. We have an incredible responsible AI team that developed strong guidelines for when and how AI can be used and specific areas where we will not use AI. All new AI use cases for our clients must be reviewed by this team. As we look at how our people are adopting AI, we also track how inclusive the usage is. Our external research shows a gap between women and men at more junior levels when it comes to genAI usage. We see similar trends across various cohorts and are working with our employee affiliation groups to help close any gaps.”

What is the one piece of advice would you offer other organizations piloting or going all in with AI?

Pittman: “Ultimately, transforming with AI should all be in pursuit of delivering value for your stakeholders. Set that as your north star and then build the governance, partnership ecosystem, tools, and new businesses you need to get there.   

“Also, this technology is fundamentally exciting and engaging. It’s about bringing more facts and ideas together, spurring creativity, freeing up time from the must-do’s that can be a drag on our time and energy. As we say, ‘less toil, more triumph.’ So, make it fun and use AI as an opportunity to unlock the creative capacity and energy of your people.”