AI Build Trap: Why Bespoke Software Is the Wrong Bet Right Now Character count

A company (maybe yours!) spends eight months building a custom digital asset management tool. Their engineers were excited. The demos were clean. Then Adobe released a feature that did 80% of what they built, natively, with enterprise-grade security and AI already baked in. Eight months of runway, gone. The custom code became a maintenance burden the moment it shipped.

That story is repeating across enterprises right now. And the AI boom is accelerating it.

The Foundation Fallacy

Here is the trap: AI has made it cheaper and faster to write software. That is real. But faster coding does not equal smarter direction. What many companies are doing with this capability is reaching for the wrong problem. They are building infrastructure when they should be building differentiation.

Call it the Foundation Fallacy. The belief that because you can build the foundation yourself, you should. In most cases, you should not.

The foundation of your digital operation, security, identity, integrations, content management, workflow automation, customer data infrastructure, is not your competitive advantage. It is the floor. And the enterprise software industry has spent three decades pouring that floor for you.

This Has Happened Before

If this moment feels familiar, it should. The 1990s internet boom triggered the exact same pattern. Every large company decided it needed to build its own web infrastructure, run its own email servers, and maintain its own intranet platforms. The reasoning was identical to what you hear today: more customizable, more controllable, more ours.

By the mid-2000s, most of those bespoke systems were liabilities. The companies that had quietly adopted Salesforce, Oracle, and SAP as their foundations were compounding. The companies still maintaining custom middleware and homegrown CRM systems were spending their engineering budget on upkeep, not growth.

The AI moment is the same bet, with higher stakes.

Boardrooms are in a version of that same freeze right now. So teams build proofs of concept, engineers prototype tools, and product managers write requirements for systems that already exist. The companies that are moving are not building from scratch. They are building on top of established platforms and competing at the layer that actually matters: the experience. (Also read: The Velocity Gap: Why AI Will Collapse Housing Before Washington Notices)

What the Best Enterprise Platforms Already Solved

The foundation layer already lives, let me break it down by domain.

Content, creative, and customer experience. Adobe owns this layer. Creative Cloud powers design and production at scale: Photoshop, Illustrator, Premiere, After Effects, and Firefly AI for generative content. Experience Cloud handles customer journeys, analytics, and personalization across every channel. Document Cloud streamlines agreements and workflows that most organizations are still running manually. Together they form a connected foundation that most companies cannot replicate internally, and would spend years trying. Adobe Experience Cloud is used by 87% of Fortune 100 companies across 12,000 enterprise customers globally (Adobe EMEA Summit, 2023), and a Forrester Consulting study commissioned by Adobe found it delivers a return on investment of over 330%. While other platforms own their domains, Adobe dominates the creative-to-customer experience layer that every modern business runs on. This is where differentiation increasingly lives.

CRM, sales, and revenue operations. Salesforce owns this layer. Two decades of investment in data models, compliance infrastructure, and integration ecosystems now powers AI agents through Salesforce Agentforce. Building a bespoke CRM in 2026 is not a competitive move. It is a tax on your engineering team.

IT, HR, and operations workflows. ServiceNow owns this layer. IT service management, HR service delivery, procurement, and operations orchestration are consolidated into a platform that handles cross-functional workflow complexity most companies badly underestimate before they start building it themselves.

ERP, finance, and supply chain. Oracle and SAP own this layer. Oracle Cloud ERP and SAP S/4HANA represent decades of accumulated domain knowledge about how money, inventory, and supply chain actually work at enterprise scale. Building your own ERP foundation because AI makes coding faster is exactly the kind of reasoning that produces eight-figure write-offs.

The Home Depot is the clearest example of the right orientation. They used Adobe Workfront to increase completed how-to DIY guides by 286% in a single year without rebuilding their creative infrastructure. They directed their engineering investment toward the experience layer: the mobile app, the augmented reality paint tool, the personalized product recommendations. Melanie Babcock, Vice President of Integrated Media at The Home Depot, said: "By unifying our data, we started waking up to the fact that our customers' trust is an extraordinarily valuable asset. They were telling us exactly what they were looking for, and we needed to be more aligned with ways to help them." (Adobe Customer Success Story)

The orientation is not "what should we build?" It is "what are our customers already telling us, and how fast can we respond?"

Even companies with highly specialized needs usually discover that only 10 to 20 percent of their requirements are truly unique. The other 80 percent is standard infrastructure that a mature platform handles better, faster, and more securely than anything built in-house.

MCP Servers: Why Non-Technical Leaders Should Care

Model Context Protocol, MCP, is the connection standard that lets an AI model reach into your actual data and systems and take action. Not generate text about your data. Actually work with it.

Think of it this way: every AI tool your company has licensed operates in a silo unless it is connected to your real systems. MCP is the USB-C moment for AI. Before USB-C, every device had a different cable. MCP is the universal connector for AI and enterprise software.

The major enterprise platforms are all moving in this direction. Adobe has baked MCP endpoints into its core platform with integrations across seven frontier AI partners including Anthropic, AWS, Google, Microsoft, and NVIDIA. Salesforce Agentforce, ServiceNow AI agents, and Oracle AI follow the same pattern. A company running on any of these platforms can instruct an AI agent to query data, identify patterns, generate content, and push it to a workflow, without custom integration pipelines. The same instruction from a company with a bespoke stack requires months of connector work first.

The March 2026 AI and Digital Trends Study found that 75% of organizations cite data integration and quality as their top AI implementation challenge. That problem shrinks significantly when your foundation is already integrated.

Build Where It Counts

The company in the opening spent eight months building a tool Adobe shipped as a feature. That is not a story about being slow. It is a story about solving the wrong problem. The engineers were talented. The execution was solid. The direction was wrong.

The foundation is not the strategy. The foundation is what lets the strategy exist. Identify which enterprise platform already owns the layer you are thinking about rebuilding. Build your differentiation on top of it, connect it with MCP, and put your engineering talent into what only your business can create.

Everything below that line is plumbing. You should not be building your own plumbing in 2026.

In the years ahead, the gap between companies with clean platform foundations and those carrying bespoke infrastructure debt will widen faster than most leadership teams expect. The compounding effect of AI on top of an integrated foundation is not linear. The companies that made the right foundation decision early will pull further ahead each quarter. The ones that did not will feel it in every sprint.

Key Takeaways

• The Foundation Fallacy is the belief that because AI makes building cheaper, companies should build their own infrastructure. The 1990s internet era proved this wrong. AI is about to prove it wrong again.
• Every major enterprise domain already has a platform that owns the foundation layer: Adobe for content and customer experience, Salesforce for CRM and revenue, ServiceNow for IT and operations, Oracle and SAP for ERP and finance.
• Adobe Experience Cloud is used by 87% of Fortune 100 companies. A Forrester study commissioned by Adobe found it delivers over 330% ROI. These numbers reflect what happens when companies stop rebuilding solved problems.
• MCP is making the integration argument stronger, not weaker. The major platforms are all publishing MCP endpoints. Building bespoke infrastructure now means building bespoke connectors too.
• The real AI advantage in 2026 goes to companies with clean platform foundations that can iterate fast at the experience layer, not to companies still in development on infrastructure someone else already solved.

About the Author

Martin Goetzinger has spent his career in enterprise software sales, helping large organizations such as Apple, Microsoft, and Verizon connect data, insight, and action. His work focuses on transforming how businesses measure success and create customer value through technology.

Outside the enterprise world, he writes about the five forces he believes are reshaping everything: AI, blockchain, energy, personalized health, and robotics. Not from a purely technical lens, but from a human one as to how these technologies will redefine work, wealth, and well-being.

He is based in the U.S. and publishes at www.MartinGoetzinger.com.

Disclaimer

The views expressed in this article are the personal opinions of the author and are provided for informational and educational purposes only. This article is not sponsored by, affiliated with, or endorsed by Adobe Inc., Salesforce, ServiceNow, Oracle, SAP, or any other company mentioned herein. All product references, case study data, and company examples are drawn from publicly available sources and do not represent any official position of those companies. Nothing in this article constitutes investment advice, financial advice, legal advice, or any other form of professional advice. Always consult a qualified professional before making decisions that affect your finances, business, or livelihood.