Tony knew it was time for an upgrade, but the question loomed large: should they build a custom CRM in-house or buy a market-leading solution like Salesforce?
As Tony sat in his office, he weighed the pros and cons of each option. Buying a packaged CRM offered an immediate, tried-and-tested solution with robust features, continuous updates, and extensive support. However, the cost was significant, and Tony was concerned about the long-term implications of relying on an external vendor. Additionally, Tony’s company had unique requirements that would demand substantial configuration changes or customization.
On the other hand, building a custom CRM presented an opportunity to create a system tailored to their exact needs. Historically, this would have been a time-consuming and expensive endeavor, but Tony was intrigued by the advancements in artificial intelligence and low-code development platforms that had emerged over the past few years. AI-powered development tools were accelerating the software development process, reducing costs, and allowing for a high degree of customization with minimal to no coding. A few years ago, he wouldn’t have imagined ever seriously considering building a CRM given the number of good options on the market.
Tony decided to explore the AI route further with his team. The technology had indeed advanced to a point where AI could handle much of the heavy lifting, from coding and testing to continuous integration and deployment. He was quite familiar with using AI to create small apps – his team recently used AI to build a time-tracking app and a customer portal – but he had never considered using it to build a large enterprise system. He discovered that AI tools had “agents” that could handle the complexity and walk through the various components of application development, including feature development, integrations, data modeling, and UI/UX. Prompt engineers could upload their existing business process documentation and provide additional requirements. If the engineer needed to change something, they could either tweak the requirement and the AI tool would rebuild it, or they could make the change in the tool’s drag-and-drop no-code interface.
With the ability to rapidly prototype and iterate, Tony’s team could potentially develop a fit-for-purpose CRM system that aligned perfectly with their workflows and growth strategy in a fraction of the time and cost that it would take to implement a packaged CRM or develop their own with a traditional approach. He wouldn’t need expensive system implementation consultants or large development teams. Much of the support and maintenance lift could be supported by AI as well.
For Tony, these AI-powered app development tools mitigated his two biggest concerns with building a solution in-house: cost and time. Tony typed up his proposal for a custom CRM built with AI-tools and sent it to the CEO. Tony then sat back in his chair and pondered if he could take a similar approach for his company’s other large enterprise systems.
—
Tony’s story is of course fictional, but it highlights a possible shift in how executives will approach their enterprise application landscape in the future with advancements in AI. The build versus buy question, which has long dominated the decision-making process for enterprise software, is likely to undergo an AI shakeup in the coming years. The general rule-of-thumb is to develop custom solutions only to support business capabilities that are unique or provide out-sized business value. For everything else it typically makes more sense to buy packaged solutions. With AI advancements though, it’s likely to be more reasonable to build more of your applications than to buy and cobble together several off-the-shelf solutions. So, is the bar to the custom path lowering?
Here, we explore how AI could reshape these software decisions, offering new opportunities and challenges for businesses.
The Traditional Approach: Where We Are Today
Today, the decision to build or buy an enterprise application hinges on a few key factors: cost, time, customization, and control. Buying off-the-shelf solutions offers immediate functionality, robust support, and regular updates, albeit at a high cost and with potential limitations in customization. Conversely, building a custom application provides tailored features and complete control but often requires substantial investment in development time, resources, and ongoing maintenance. Many companies do not have the product and development talent to pull it off successfully either.
There are nuances and exceptions, but custom development, particularly for large enterprise applications, is typically a massive effort, fraught with risk, and requires substantial maintenance. Implementing packaged solutions is certainly no small task either, but it’s typically an easier path to the functionality a company needs. Today in 2024, Tony would probably recommend buying a CRM, not developing a custom CRM.
AI-Driven Approach: The Future?
Nowadays, it is hard to go a day without hearing about AI and its potential. A tremendous amount of investment is pouring into this exciting new suite of technologies, yet most people’s only hands-on experience with AI is through ChatGPT or other productivity-focused tools. It’s the early days for all of us, and it’s fair to be skeptical of the hype and promise of AI. Perhaps the scenario described in Tony’s story seems too far-fetched and impossible to you. Maybe, but several interesting developments are happening today that suggest this future may be right around the corner:
- Software engineers already use AI tools like Github’s Copilot and Cursor to accelerate development by automating repetitive tasks, providing code suggestions, and real-time error detection. A Gartner survey showed that 63% of organizations were using or piloting AI code assistants in 2023, and the number is likely higher today.
- Amazon AWS recently released a preview of their App Studio, a gen-AI service that allows individuals to “create enterprise applications in minutes.”
- Amazon CEO, Andy Jassy, reported that efficiencies from their coding assistant, Amazon Q, resulted in $260M in savings or 4,500 developer years!
- Microsoft is baking AI-powered development into their Power Apps along with no- and low-code interfaces, allowing individuals to deliver custom apps quicker.
- Financial institutions are using Flowx.ai to modernize and transform their application stack in a matter of weeks.
- Codium AI, considered more of an AI code assistant today, is actively working toward an AI agent approach to building enterprise applications.
- Klarna, the fintech platform, recently made the news for trying to replace their Salesforce and Workday applications with in-house, AI-built applications
- To date, GenAI models have been limited in their ability to do math and solve logic problems as they’ve been trained on large language learning models and linguistic patterns. But that is changing. For example, OpenAI’s new model, Strawberry, scored 90% on the MATH benchmark.
There are plenty more examples and it remains to be seen whether these tools will fully deliver on their promise, but with enough time it seems inevitable. Companies are already using or exploring AI code assistants and AI-powered app development tools to accelerate development efforts and create small bespoke applications. It’s reasonable to think the next evolution will be addressing capabilities typically delivered by large enterprise applications like ERP, CRM, PLM, MES, HRIS, etc. We are already seeing signs of companies looking at this next step. For example, Klarna, the fintech company, recently made the news by trying to replace their Salesforce and Workday applications with in-house AI-built applications. This is the world we envision in Tony’s story, and it might arrive much sooner than 2030.
Potential Benefits of an AI-Driven Approach to the Application Stack
Artificial intelligence, particularly in the form of low-code and no-code platforms, will revolutionize how businesses approach the build versus buy decision. Here are some of the key impacts of AI on this process:
- Accelerated development cycles: AI-powered development tools significantly reduce the time needed to build custom applications. Automated coding, testing, and deployment capabilities allow businesses to develop and launch applications much faster than traditional methods. This acceleration not only saves time but also enables companies to respond more swiftly to market changes and customer needs. Companies may be able to create complex application prototypes in a matter of days in the future.
- Cost efficiency: By automating many aspects of the development process, AI reduces the need for extensive human labor, which traditionally drove up costs. Businesses can now distribute resources more efficiently, directing savings toward other strategic initiatives. Furthermore, the reduced time to market can lead to quicker returns on investment.
- Enhanced customization: AI enables a higher degree of customization without the complexities associated with manual coding. Advanced AI algorithms can analyze business processes and suggest tailored features that align closely with organizational needs. This ability to fine-tune applications to specific requirements enhances operational efficiency and user satisfaction.
- Scalability and maintenance: AI-driven applications can adapt and scale more effectively. Predictive analytics and machine learning models can anticipate future demands and adjust resources, accordingly, ensuring consistent performance. Additionally, AI can automate routine maintenance tasks, reducing downtime and freeing up IT teams to focus on more strategic activities.
- Risk mitigation: Building custom applications traditionally carried significant risks, including project overruns, technical debt, and security vulnerabilities. AI could mitigate these risks through automated quality assurance, continuous monitoring, and advanced security protocols. These features ensure that applications are robust, secure, and compliant with industry standards.
Challenges and Considerations of an AI Approach
While AI-powered development can offer many advantages, it also presents new challenges that businesses will need to navigate:
- Initial investment and ongoing costs of AI tools: The adoption of AI-powered development platforms will still require an upfront investment in tools and training, plus ongoing costs. Businesses must still weigh these costs against the long-term benefits and savings.
- Skill requirements: While AI reduces the need for extensive coding expertise, there is still a need for skilled professionals who understand AI technologies and can manage AI-driven projects effectively. Upskilling the workforce or hiring new talent may be necessary. For example, “prompt engineering” will be a key skill in this AI-dominated world.
- Data and integration with existing systems: Integrations and data management are likely to be key limitations in the first iteration of AI app-building tools that will still require hands-on work to create and manage. Integrating AI-built applications with legacy systems can be complex. Businesses need to ensure seamless interoperability to avoid disruptions and maximize the benefits of their new solutions. Converting data from legacy systems could still be a highly painful process. Future iterations of AI tools will hopefully help address these challenges.
- Privacy and security: AI systems often rely on large datasets to function effectively. Ensuring data privacy and security is paramount, especially in an era of increasing regulatory scrutiny and cyber threats. Implementing robust AI-specific governance frameworks and standardized operating procedures (SOPs) can help organizations proactively address risks and ensure compliance with evolving regulations.
- Process Management: If a company has poor process definition or discipline, those problems will affect any AI-developed application. At the end of the day, a company will still have to have a solid grasp of what they do, how they do it, and how technology will enable it.
Finally, while not a new challenge, AI-powered development for custom software increases the vital importance of a product-focused mindset and surrounding discipline.
Naturally, there are other factors that will affect the trajectory of AI-powered app development. For one, software companies are not going to sit idly by as AI-powered development becomes more prevalent. Many are already embedding AI into their existing products. They will do everything they can to keep their software relevant with proprietary features, and they may be successful or a victim of disruption. However, how AI will impact the software industry is worthy of its own article.
Conclusion
The future of enterprise application development is undoubtedly intertwined with AI, offering exciting possibilities for organizations willing to embrace this technology. Leveraging AI could provide the best of both worlds, customization and cost-efficiency, while mitigating many traditional risks associated with building custom software. No longer would building custom enterprise applications be as daunting a task as it is today, marking a fundamental shift in how executives would approach build vs. buy decisions for their application stacks.
We expect AI to empower every company to build custom software in the coming years. As the momentum builds, we recommend that companies ease their way into AI-powered custom development by starting with smaller, less business-critical applications and/or starting in areas where there isn’t good fit COTS software. Doing so would be a low-risk way to build skills and test their way into this model before tackling enterprise, business-critical software.
Looking ahead, if AI-powered development can deliver on its vision, it will soon become commonplace for companies to use these tools to rapidly build bespoke enterprise solutions across their businesses. While the initial iterations may not be practical for addressing large enterprise applications, like Tony’s CRM dilemma in our fictional story, it’s not unreasonable to envision this evolution in the future.
The implications are profound. Companies could build applications perfectly tailored to their operations in a fraction of the time it takes today. Additionally, they could easily modify their apps to meet changing business needs or experiment with innovative solutions that they otherwise would not pursue. Several trade-offs that executives weigh today when debating whether to build or buy applications may no longer be as significant with advancements in AI-powered app development.
At Cuesta Partners, we have helped many CIOs like Tony to tackle the build versus buy dilemma. With the impending AI shakeup on the horizon, speak to one of our experts today to understand how your organization can reap the benefits of bespoke enterprise solutions.
References:
Github Copilot: https://github.com/features/copilot
Cursor: https://www.cursor.com/
AWS App Studio: https://aws.amazon.com/appstudio/
Microsoft Power Apps: https://www.microsoft.com/en-us/power-platform/products/power-apps
Klarna Story: https://www.inc.com/sam-blum/klarna-plans-to-shut-down-saas-providers-and-replace-them-with-ai.html
