Saved To My Saved Content
Download Article

Companies are increasingly looking to GenAI tools to supercharge their innovation efforts. By deploying tools that make coding more accessible, companies are putting the ability to develop workflow automations or make product tweaks within reach of nontechnical employees.

A bottom-up, employee-driven approach to innovation holds significant promise. Breaking down the traditional barriers between daily execution and the exploration of new solutions promises to generate more improvements, faster. But it will take more than GenAI tools alone to truly unleash firms’ innovative potential—and enable firms to adapt to today’s volatile environment. Here’s what firms must do.

BCG Henderson Institute Newsletter: Insights that are shaping business thinking.

Emulating Nature’s System for Continuous Adaptation

Evolution is nature’s mechanism for continuous adaptation. It builds on the variation that occurs naturally across a population. Similarly, GenAI tools can increase the variation of new ideas and solutions that are developed across a corporate workforce. For example, at Chevron, employees have used low-code tools to create an app to track and manage drilling equipment availability, and a compliance audit checklist app for field inspections. Some improvements lie beyond the capabilities of the average employee—be it due to a lack of expertise, resources, or time. In these cases, GenAI can enhance the development of creative ideas that fuel innovation and complement the work of traditional R&D departments.

However, evolution complements variation with selection (of the traits with the best fit for a given environment) and amplification (the spreading of these traits across the broader population). To emulate this biological process, businesses must complement the deployment of GenAI tools with systems for testing, identifying, and rolling out the best decentrally developed solutions.

Empowering Everyone to Test and Learn

When generating and implementing new ideas is part of everyone’s job description, organizations need to put a robust testing process in place to see what works. Speed and simplicity are crucial because employees will be discouraged from contributing to innovations if they face difficulties in getting new solutions approved.

One possibility is for leaders to implement a decentralized testing system. This could take the form of granting autonomy to teams or squads to run their own tests, for example, of automated workflows or tweaked customer offerings. At Booking.com, every team can run experiments on the main website to test new features or changes to the site design.

Testing capabilities can be embedded directly in the tools that employees use to create new innovations. At PG&E, for instance, the low-code platform includes access to a managed sandbox environment so that nontechnical developers can test apps and flows safely before going live.

Another approach is to establish predefined thresholds below which central approval is not needed. This could be the case for changes that are limited in scope (such as automating a team’s internal process) or low risk (because they can be easily rolled back). For example, Shopify put in place guardrails so that new features being tested have a limited “blast radius.” They are first rolled out to only a small percentage of customers and can be quickly reversed with a “kill-switch” that either developers or customers can trigger.1 1 Anthony Cameron, “Using Betas to Deploy New Features Safely,” Shopify blog, February 12, 2021.

What about cases where employees use GenAI to develop or refine an idea? In these situations, leaders can speed up testing of early-stage ideas by setting up an internal marketplace where workers can post their ideas and invite specialists—such as the owners or tech leads of a related product—to co-create a prototype.

For example, Bayer’s platform WeSolve enables employees at all levels to submit, develop, and collaborate on ideas, whether they are related to company-wide strategic priorities or to their department. Employees engage on the platform, and dedicated innovation “coaches” are on hand to run design-thinking sessions or shepherd experiments, giving grassroots idea-owners access to domain specialists.2 2 Ben M. Bensaou, “Middle Managers: The Forgotten Heroes of Innovation,” INSEAD Knowledge website, January 12, 2022;Monika Lessl, Henning Trill and Julian Birkinshaw, “Fostering Employee Innovation at a 150-Year-Old Company,Harvard Business Review website, December 17, 2018; A similar approach is used by Intuit, where “innovation catalysts” guide colleagues through turning ideas into structured test cases and experimentation sprints.3 3 Roger L. Martin, “The Innovation Catalysts,”Harvard Business Review, June 2011.

Amplify Successful Ideas Throughout the Organization

Across all mechanisms of testing, it is crucial for firms to track and record results and impact through a central system. This is fundamental to enabling successful ideas to proliferate across the wider firm, multiplying their impact.

For example, Booking.com’s GenAI studio—which enables nontechnical staff to design and run experiments, such as alterations to its website—automatically logs every experiment.4 4 Raphael Lopez Kaufman, Jegar Pitchforth, and Lukas Vermeer, “Democratizing Online Controlled Experiments at Booking.com,” arXiv,2017. This is done alongside recording outcomes, such as impact on customer engagement. Meanwhile, leading consumer packaged goods companies, which use GenAI-powered synthetic consumer panels to conduct hundreds of automated focus group tests, feed the results straight into a central ledger. This provides teams with up-to-date insights into which formulations are likely to resonate with key segments. Finally, at Walmart’s Intelligent Retail Lab, store-level tests are tied to operational KPIs and centralized for broader rollout decisions.5 5 Matt Smith, “Walmart’s New Intelligent Retail Lab Shows a Glimpse into the Future of Retail, IRL,” Walmart website, April 25, 2019.

Some companies may choose to build on their established knowledge sharing systems, to which new solutions and workflows can be submitted, alongside descriptions of where and how they may be applied. The experiments conducted in Booking.com’s GenAI studio, alongside their outcomes, can be browsed by every user—enabling employees to see what worked and inspiring ideas on what else to try.

To allow for the diffusion of software-based solutions, these systems should be complemented with widely accessible code repositories, to which code snippets or prompts are submitted, becoming reusable across the firm.

Companies can explore a variety of paths to incentivize the sharing of new ideas and solutions. For example, leaderboards can be put in place that display the popularity of ideas (based on community votes), track the adoption count of new solutions, or even display the impact on operational KPIs across the firm. These metrics could be tied to rewards—be they monetary incentives or recognition in town hall meetings.

Another approach is to set up a dedicated knowledge curator  role—tasked with curating distributed innovation efforts—to help the organization learn from what has been tried, what worked, and what has not. For example, at NASA, “knowledge officers” are deployed to each center and directed to capture lessons learned and curate results. To disseminate the acquired knowledge, these knowledge officers form a cross-agency network that regularly convenes to exchange insights and brief NASA leadership.

What about decentrally developed ideas that go beyond small-scale operational improvements and hold promise to fundamentally shift the organization’s operations or its offerings? Realizing the value of these ideas requires ownership by a traditional R&D organization, which acts as a “scaler.”

For process improvements, such as workflow automations, R&D experts can be tasked with standardizing team-specific innovations into enterprise-grade solutions deployable across the firm. For new customer-facing offerings, companies may set up an internal accelerator. The Swedish telecoms company Ericsson, for example, has formed a dedicated team that identifies promising employee pitches, assigns a team of R&D specialists, funds rapid prototyping sprints, and shepherds those projects into the company’s growth portfolio.6 6 Daniel Alexus,“Unleashing our Superpower: Ericsson’s Bold Bet on Intrapreneurship,” Ericsson website, July 6, 2023.

A crucial prerequisite is to put in place a flexible resource allocation system—and regular review process—to ensure that, as the environment changes, capital and time are quickly shifted toward promising ideas. This could entail defining a set of leading metrics tied to the potential of a given idea. For example, food-packaging and food-processing multinational Tetra Pak tracks a multitude of signposts, such as the percentage of packaging required by law to be recycled in key geographies. Depending on how these indicators move, management launches new innovation programs, accelerates existing ones, and decides which ones to put on hold.

When Innovation Is Everyone’s Responsibility

A final requirement to unleashing employee-driven innovation is to foster a culture that emphasizes the value of innovation and the role of each employee in driving it.

This requires challenging employees’ view of the firm as a machine—efficiently executing a given task in a stable but rigid environment—and reimagining it as a living organism—continuously adapting to changing conditions. The involvement of all employees in innovation should be framed as key to such adaptation. That’s because high participation creates the necessary pool of diverse ideas from which the best can be selected and amplified.

To reinforce that each employee is responsible for developing new ideas, leaders should highlight the powerful advantage that traditional “execution” roles—such as sales—have in generating insights. Frontline employees like sales representatives are ideally positioned to identify frustrating processes or emerging customer demands. Their observations are the raw sensory data the firm needs before it can adapt. Thus, these employees should be encouraged to harness each iteration of their day-to-day tasks as inspiration for the development of new ideas.

Finally, implementing a system of employee-driven innovation does not mean that classical innovation-focused roles in engineering or R&D departments are no longer needed. Rather, the focus may shift toward roles such as “knowledge curators” and “system builders,”  who design and maintain the infrastructure (platforms, governance, and experimentation environments) that allows decentralized innovation to function smoothly and safely.


GenAI tools such as low-code platforms can empower each employee to go beyond their “day job” and either develop or even directly implement innovations. This can unlock a greater variation of ideas. When paired with new systems for selecting and amplifying the highest-potential offerings, these ideas can be a boon for companies operating amid volatility.