Adopting Artificial Intelligence for New Product Development: The RAPID Process
Dr. Robert Cooper
kHub Post Date: August 20, 2024
Read Time: 8 Minutes
AI Promises Significant Payoffs
Leading early adopters of AI in new product development (NPD) like GE and Nestlé report remarkable improvements, including over 50% reduction in development times and enhanced innovation rates. Despite these benefits, AI adoption for NPD remains sluggish, particularly among U.S. firms. A McKinsey study showed that globally only 13% of firms had adopted AI for NPD by early 2023;[1] that number had risen to 32% of German firms but only 18% of US firms by early 2024.[2] Chinese and Indian firms are somewhat ahead, with adoption rates of AI for NPD estimated at 32% and 37% of firms.[3]
AI Projects Fail to Deliver
Several factors contribute to low AI adoption rates, including lack of trust, insufficient senior management commitment, and ethical and cyber-security concerns. A major barrier is the lack of a robust business case.[4] Many AI projects struggle to demonstrate clear business value:[5] 89% of large companies had AI initiatives underway, but only 31% of expected revenue increases and 25% of expected cost savings were achieved.[6] Similarly, a Deloitte study found only 18-36% of organizations achieved expected benefits from AI.3 Despite claims of AI’s positive impact on NPD, “hard evidence that directly ties digital and AI transformation to improvements in operational KPIs and financial performance is scant”.[7]
These poor business values occur because some AI projects are just inconsequential applications – a bad choice of project – but many AI projects don’t deliver value because they simply fail in execution. Only 53% of AI projects make it from prototype to production![8] This "pilot paralysis" phenomenon, where companies undertake successful pilot projects with AI but struggle to scale up, is epidemic.[9] Some estimates place the AI failure rate as high as 80%, almost double the failure rate of corporate IT projects a decade ago.[10]
An extensive review of the available evidence identifies the main reasons for AI project failure.[11] Many are familiar to experienced product developers, including a failure to understand the users’ needs, and inadequate product testing.
A Map for the AI Journey
AI projects share many similarities with B2B new product projects. In both, the ultimate goal is to deliver a solution that meets the needs of business end-users, whether they are internal people (for AI projects) or external customers (for NPD projects). The tasks and stages in AI and NPD projects are also similar, such as building a business case, doing product tests, and going to full production. One major equipment business simply modified their NPD gating process and used it for their internal AI projects.
Implementing AI is a complex journey with obstacles and pitfalls along the way. Setting out on such a long journey without a map is unwise. The deployment of new technologies and new ways of working is not new to NPD and RD&E people, however. Numerous frameworks for technology development and acquisition have been successfully used for almost three decades.
The first technology development (TD) model was pioneered by Thamhain in 1997, modified from the original Stage-Gate® process.[12] Shortly after, Cohen and colleagues developed a similar modified gated model for chemical technology developments at ExxonMobil Chemical.[13] The project Metallocene (now a billion-dollar business) was one of the first projects through that system! Cooper’s detailed Stage-Gate-TD model was modeled on that ExxonMobil system.[14] A summary TD model appeared in the 2002 PDMA Handbook.[15] And Lercher’s "Big Picture Model" was subsequently created to handle all types of innovation projects, including TD projects (see Figure 1).[16]
Figure 1. The “BIG PICTURE” Innovation & Technology Development Process
These models work! A study of six firms that implemented such TD models helped validate the approach and concluded that companies must create a more flexible approach for technology projects compared to NPD.[17] The TD model should incorporate "iterative loopbacks, recursive use of the model, redefining development goals and objectives based on generated new knowledge, and flexibility in timing."
A number of firms, besides ExxonMobil Chemical, have developed their own TD models. For example, 3M employs a three-stage TD process, called NTI, that creates new technologies (Figure 2), which feeds their five-stage new product gating process, NPI.[18]
The RAPID Process Map
Our “Roadmap for AI Procurement, Implementation, and Deployment” (aka RAPID in Figure 3) is based on these proven TD models as well as on best practices in AI deployment. It is a stage-wise process with incremental but increasing commitments of resources at each stage, thus is a risk mitigating model. The model addresses common reasons for AI project failure, adoption hesitation, and major risks, providing actions and best practices to avoid these pitfall
Figure 2: 3M’s Two Innovation Processes – NPI for New Products, & NTI for Technology Development (TD).
Figure 3. The RAPID Technology Acquisition & Deployment Process Map for AI in NPD
Main Features of RAPID
- Four main stages and gates: RAPID includes stages where the AI project team undertakes specific tasks, with management engagement and signoffs at gates to ensure commitment and resource allocation.
- Decision points: Gates also serve as "go/no-go" decision points to kill weak AI projects and pivot to better applications.
- User needs focus: Employing VoP (voice of process) and VoB (voice of business) ensures alignment with user needs, avoiding the "solution looking for a problem" scenario.
- Robust front-end homework: This includes a robust technical assessment, vetting AI vendors (Table 1), and building a fact-based business case.
- Iterative approach: Early stages incorporate build-and-test iterations, borrowed from Agile Development methodology, to allow experimentation and fixes.
- A tough Pilot stage: The selected AI solution is piloted with close monitoring for necessary improvements, a critical stage often beset by problems.
Stage-by-Stage Walkthrough
Let’s do a walkthrough RAPID, using Figure 3 as a guide; more details on each stage are provided in Figures 4, 6 and 7.
Vision and Purpose: Deploying AI in NPD must be a senior-management-sponsored initiative. The leadership must commit to the AI initiative, agreeing on its vision, purpose, and goals. A cross-functional Task Force or AI Ops Team is created, reporting directly to senior management, ideally composed of dedicated members and responsible for driving the AI initiative forward. External expertise may be added to provide necessary knowledge and skills.
Once the team is in place, management meets with the Task Force to finalize and agree on the next steps. At Gate 1, management signs off and commits the needed resources for Stage 1 – see Figure 4.
Figure 4. Key Tasks in the Early Stages of RAPID – Vision & Business Case
Stage 1 – Build Business Case: The Task Force gets up to speed on AI for NPD and undertakes critical homework tasks, including researching user needs, vetting AI solutions (Table 1), and determining the business value of relevant applications (see Figure 4). Deliverables include summaries of recommended AI applications, their business cases, and an AI Application Projects Roadmap as in Figure 5.
Figure 5. The Application Projects Roadmap is a tentative plan for future AI application projects
Gate 2 approves the proposed AI project(s) and commits resources for acquiring or developing the solution. If the decision is to develop the solution internally (that’s rare), the project becomes an internal IT development project and reverts to that IT process, usually an Agile-Scrum method.
Stage 3 – Solution Acquisition: The AI solution is acquired, either purchased or leased, and Proof of Concept (PoC) tests are undertaken. Tasks include software installation, training, and PoC testing to ensure the solution works and integrates seamlessly with current systems – see Figure 6.
Figure 6. Key Tasks in Stages 2 & 3 – Solution Acquisition & Pilots
PoC testing is critical to validate the AI solution’s performance and compatibility with existing systems. Too many projects reach the next stage, Pilots, only to discover major problems. The Task Force should plan and execute a series of test-and fix cycles or demo iterations to gauge technical performance and demos to validate stakeholder acceptance in Stage 2, well before piloting. Deliverables to Gate 3 include an Action Plan for the Pilots, an updated business case, and an Application Projects Roadmap. Gate 3 approves the project for piloting and commits the needed resources.
Stage 3 – Pilots: The Pilot Stage tests and validates the AI solution with real users and under real operating conditions. The project team implements the Pilot Plan, including training, communication, providing support, monitoring, and measuring results – see Figure 6. They gauge the impact of the solution on performance, how well the product integrates with the existing system or process, and measure users’ satisfaction and intent-to-use in the future. The business case is updated, and a Plan of Action is created for the next stage, Scale to Production. Gate 4 releases the solution to production or full operation, approves the Scale-up Plan, and commits the resources for Scaling
Stage 4 – Scale to Production: Scaling involves moving the AI solution from a controlled pilot environment to full production, ensuring it can handle real-world data volumes, operate reliably, and integrate with existing systems. Key tasks are those typical of a change management program, namely communications, training, installation and start-up – see Figure 7. Once the AI is in full production, then follows monitoring, measuring performance, and continuous improvement.
Figure 7. Key Tasks in the Scale-Up Stage of RAPID
Conclusion
Time is running out for firms yet to embark on the AI journey, but it’s not too late to start. Senior management must commit to an AI initiative, establish a dedicated Task Force, and follow a map like the RAPID model to guide the path forward. AI is poised to power companies and industries, offering a competitive advantage to early adopters. Now is the time to push ahead on your "AI in NPD" journey.
The RAPID model offers a comprehensive roadmap for AI procurement and deployment. It emphasizes the importance of understanding user needs, conducting robust front-end homework, and incorporating iterative testing cycles. By adhering to this model, firms can successfully navigate the complexities of AI deployment and avoid the common pitfalls along the way.
References
[2] Robert G. Cooper and Alexander M. Brem, “The Adoption of AI in New Product Development: Results of a Multi-firm Study in the US and Europe,” Research-Technology Management 67(3), (2024)33—54: (2024). DOI: 10.1080/08956308.2024.2324241
[4] Robert G. Cooper, “Overcoming Roadblocks to AI Adoption in New Product Development,” Research-Technology Management (forthcoming in Sept. 2024).
[12] H.J. Thamhain, “Using Stage-Gate Processes Effectively for Managing Technology-based Projects,” Innovation in Technology Management. The Key to Global Leadership. PICMET ‘97, Portland, OR, USA (1997): 398, doi: 10.1109/PICMET.1997.653441
[13] L.Y. Cohen, P.W. Kamienski, and R.L. Espino, “Gate System Focuses Industrial Basic Research,” Research-Technology Management, 41(4) (1998): 34–37.
[14] Robert G. Cooper, “Managing Technology Development Projects,” Research-Technology Management, 49(6) (2006): 23–31.
[15] P.A. Koen, G.M. Ajamian, S. Boyce, A. Clamen, E. Fisher, S. Fountoulakis, A. Johnson, P. Puri, P., and R. Seibert, R. “Fuzzy Front End: Effective Methods, Tools, and Techniques,” The PDMA Toolbook for New Product Development, Chapter 1 (2002): 2–35. John Wiley & Sons, New York, NY.
[16] H. Lercher, “Big Picture – The Graz Innovation Model,” CAMPUS 02, University of Applied Sciences, Germany, (May 11, 2017, last revised May 20, 2021). Link: https://ssrn.com/abstract=2965373
[17] Hans Högman and Ulf J. Johannesson, “Applying Stage-Gate Processes to Technology Development—Experience From Six Hardware-Oriented Companies,” Journal of Engineering Technology Management, 30, (2013): 264–287. https://doi.org/10.1016/j.jengtecman.2013.05.002
[18] Tom Gehring, “Sustaining an Innovative Culture at 3M,” Proceedings, Stage-Gate International SUMMIT, Miami, FL. (2011).
About the Author
Dr. Robert Cooper,
Professor Emeritus, McMaster University, Canada
ISBM Distinguished Research Fellow at Penn State University
A world expert in the field of management of new-product development and product innovation, Dr. Cooper has written 10 books on the topic and more than 150 articles. Bob is the creator of the globally-employed Stage-Gate (trademarked) process used to drive new products to market; a Fellow of the Product Development & Management Association; ISBM Distinguished Research Fellow at Penn State University. He is a noted consultant and advisor to Fortune 500 firms, and also gives public and in-house seminars globally.
LinkedIn