Execution Is Abundant. Integration Is Not.
Building software has become easy. The real challenge is connecting everything. Successful companies focus on integration and demonstrate its importance for digital projects.
Gartner reports that only about half of digital projects met their goals in 2024. Ongoing volatility is making this worse. In 2025, changes in U.S. tariffs forced companies to rework their supply chains quickly. Major automakers switched most of their contracts to overseas suppliers almost overnight. A major cyberattack in healthcare created a large claims backlog and disrupted cash flow for many U.S. providers. With ongoing geopolitical tensions, wars, and supply chain shocks, what worked six months ago might not work now. This means digital projects need a new approach: test quickly, show value within weeks, and only expand what proves itself.
Meanwhile, another shift happened. Between 2020 and 2024, building software became much faster and cheaper. Low-code platforms became common, and Forrester says 87 percent of enterprise developers now use them. AI sped things up even more. A central bank study found that developers using coding assistants wrote about 55% more code, especially junior engineers. Reuters reported that by late 2025, almost half of all startup funding worldwide went to AI products that speed up coding, testing, and software release.
There is a clear clash: on one side, volatility; on the other, faster building with low-code and AI. In between, integration must connect new systems without causing problems. This integration layer needs to handle disruptions and keep the business running. Teams know how to build, but the tough questions are what to build, how it fits with current systems, and if it will keep working as things change.
A different starting point
What if you focused on the flow of value instead of just listing features? Choose one key process, like order-to-delivery. Set a few simple rules for each step, such as “update inventory within two minutes,” and test the entire process daily. Design for real situations, not just a list of features. If demand rises by 15 percent, do prices update, do allocations adjust, and can you still keep your promises to customers? Test this in real operations, not just in demos. Run end-to-end tests, look for weak spots, and set a few clear goals to track each week. Roll out changes to a small group first and make it easy to reverse them if needed. Assign a small platform team to manage handoffs and keep things running smoothly. Before any change goes live, run a full ‘day in the factory’ simulation.
Integration in practice
This same approach works in manufacturing. Modern supply chains must detect problems early and respond quickly: catch fraud before it happens, predict machine issues before the line stops, and reroute shipments when risks increase. When rules are clear, AI can handle routine decisions like setting reorder points, adjusting schedules, and organizing trucks. People remain responsible, and decisions are made faster.
The real breakthrough is in how everything connects. The AI model must link with procurement, inventory, scheduling, customer promises, finance, and compliance so the whole system works together. For example, a Midwest automotive supplier connected predictive maintenance with procurement and scheduling using shared event streams and clear performance agreements. They also set up automated gates to block risky releases. As a result, their order-to-promise time dropped by 28 percent, and their change-fail rate went from 9 percent to 3 percent.
This pattern is true in many industries. As AI speeds up building, integration becomes even more important. In 2024, a large platform team used text-to-workflow tools to create service automations, then improved them with contract checks and ‘day in the life’ tests. Within weeks, about half of the new workflows were auto-generated and safely used in HR, finance, and IT.
Looking ahead to 2026, it helps to use a mental model that starts with value and works backward: Build, Integrate, Operate, Value. Build is about creating features and models, which is now easy to do. Integrate means connecting them to real data, controls, and teams. Operate is about running them reliably in real situations. Value comes when this flow leads to real business results. Most digital projects fail between Build and Integrate—not because the code is wrong, but because it does not connect well with the rest of the business. That is the gap to fix.
The AI acceleration question
As AI speeds things up, it also raises harder questions. Where can agents add value without increasing risk? This works best where rules are clear and any possible harm is limited. What makes autonomy safe? You need clear agreements on inputs and outputs, easy-to-explain monitoring, the ability to reverse actions, and backup plans if things go wrong. AI speeds up execution, but whether that leads to value or chaos depends on how well you handle integration.
The real choice is not just between going faster or slower. Instead, ask where the bottleneck is and if your spending matches it. Companies that focus only on building quickly may see small wins but also create bigger problems elsewhere. Those who focus on delivering value quickly treat integration as a product and build up small successes into lasting advantages.
If you already have strong execution, where should you invest next to get lasting reliability?
Disclaimer: The authors are writing in their personal capacity. The views expressed in this article are their own and do not represent the positions of their employers.
About the authors
Madan Ramachandran is a senior business leader with 20+ years of experience driving growth, scale, and transformation at SAP, Qlik, Salesforce, and Amazon Web Services. He focuses on translating vision into operating mechanisms, go-to-market strategies, and measurable outcomes for customers. He writes in his personal capacity. Follow on Linkedin for future updates
Dr. Sachin Sharma is an enterprise supply chain leader with 20+ years of experience implementing ERP and supply-chain programs across complex enterprises. He researches agentic systems and integration patterns that connect decisions to operational reality. He writes in his personal capacity. Follow on LinkedIn for future updates.