Opinion
The strategic mistake most managers are making right now with AI
"Managers who continue to see AI as just a tool for improving coding productivity will be left behind. The real winners will be those who understand they have been given a "universal brain" that can learn and improve every process in the organization," writes Asaf Amran, CTO at Sela.
Ever since GenAI burst onto the scene, most discussions in the tech world have focused on AI's ability to generate code. Many executives are considering or have already begun adopting these tools, based on the perception that the primary benefit is improving developer productivity: they will write code faster, and we will deliver more features in less time. While this perspective is important, it represents a costly strategic mistake.
Focusing exclusively on code generation is like buying a sophisticated smartphone just to use the calculator. Managers who measure AI's success in their organization solely by "lines of code" or "development speed" are missing the real revolution: AI's ability to transform and upgrade the entire end-to-end (E2E) software development process, and subsequently, every business process in the organization.
The true value lies not just in performing a single task faster, but in using AI as an autonomous agent capable of managing entire processes. In the workshops I conduct for managers and developers, we demonstrate this exact process. Instead of starting with writing code, we start with a vague business idea.
Instead of a whole team wasting hours in discussions, we ask AI to distill the idea into a technical specification, write clear User Stories, and even format them as a CSV file to automatically upload them as tickets to our Jira project. Before a single line of code is written, we have already saved days of planning and coordination.
From there, we move to the development environment (IDE). An AI agent that has indexed all our code repositories understands the project's context and is able to build the application skeleton for us, write the business logic, and, importantly, also write the Unit Tests for that code.
This capability peaks when bugs are opened in Jira (here too, the QA team uses AI to open the bugs). The agent identifies the bug, reads the description, locates the bug in the code, suggests a fix, writes a test to ensure it doesn't regress, performs a Push to GitHub, and updates the Jira ticket to "Done" status, including a detailed description.
This is no longer just "writing code"; it is the automation of entire processes. This concept, using AI to manage end-to-end processes, is not limited to the development department at all. A manager who sees this potential will immediately understand that it can be applied everywhere.
Instead of asking the finance department for a quarterly expense report, an AI agent can be activated to analyze all expenses against forecasts, identify anomalies, and write an executive summary with actionable recommendations for the CFO. Instead of the procurement department manually reviewing contracts, an AI agent can scan all supplier contracts, compare terms and prices, and automatically flag contracts that are approaching their renewal date.
Here we arrive at the most important point, the new mindset that managers and organizations must adopt: "AI-First" thinking. My message to everyone I meet is simple: For every task before you, big or small, stop and first ask yourself: "How can I use AI to perform this?" This applies to every employee, in every department.
Before an important meeting with management, instead of opening a blank presentation, ask AI: "Analyze my team's last five reports and the CEO's last three emails, formulate three key points I must present in the meeting, and also note what questions my manager might ask me during it."
Managers who continue to see AI as just a tool for improving coding productivity will be left behind. The real winners will be those who understand they have been given a "universal brain" that can learn and improve every process in the organization. The question is no longer if AI can help, but how we harness it to rethink everything we do.
Asaf Amran is the CTO at Sela.
