Endava Applied AI Agents to Reduce Requirements Analysis from Weeks to Hours

Endava has shared its experience in transforming its organization using AI agents based on OpenAI's Codex. The main result of implementation — project requirements analysis now takes hours instead of weeks, significantly accelerating the entire development cycle.

What is an agentic organization

An agentic organization is a company that integrates AI agents into workflows so that they take on routine, mechanical tasks. This frees people for work that requires judgment, creativity, and strategic thinking. Endava applied this model to one of the most critical stages of software development — requirements analysis. This is the stage where time is often lost, misunderstandings arise between client and developer, and errors accumulate. AI agents in this process work as an analyst's assistant: they study source data, highlight the main points, propose structure, and check logic. The human remains in the role of an expert who makes final decisions.

From weeks to hours

The classical requirements analysis process is a multi-stage pipeline:

• Collecting initial information from the client (often in the form of emails, documents, meetings)
• Documenting requirements in structured form
• Checking requirements for internal contradictions
• Clarifying disputed points with the customer and internal team
• Aligning requirements with project architecture
• Preparing final specifications for developers

This work requires concentration, attention to detail, but does not require creative problem-solving. This is exactly what Endava entrusted to agents. As a result, the company reduces this phase from weeks to hours. The agent quickly structures chaotic information, identifies gaps (which a human might not notice the first time), and proposes specification options.

Where the saved time goes

The main idea of an agentic organization is not to lay off analysts, but to redirect them toward more valuable work. The freed-up time an analyst spends on:

• In-depth conversations with the client (instead of mechanically rewriting information)
• Verifying agent hypotheses and correcting its errors
• Architectural consulting on agent proposals
• Managing project risks and planning reserves
• Coordinating between different project teams

This is work that requires experience, business acumen, and where a human creates real value.

Effect on the entire development cycle

Reducing the requirements phase is the first domino. When requirements are ready faster and more accurately, the entire rest of the cycle accelerates: sprint planning, development, testing, delivery. For Endava, which works on contract development, every day of acceleration is a competitive advantage and increases project margins.

What this means

Endava demonstrates that agentic workflows are moving from experiments into production practice. For the industry, this is a signal: companies that adopt AI agents early in projects will be able to work faster, more accurately, and bring products to market ahead of competitors.