Postman survey finds increased reliance on APIs to drive revenue but many of those APIs may not be ready for the AI era.
A survey of 5,751 developers and technology professionals published this week finds nearly two-thirds (65%) work for organizations generating revenue via application programming interfaces (APIs) they have deployed.
Conducted by Postman, a provider of a platform for building and deploying APIs, the survey also finds 46% of respondents work for organizations that plan to spend more time and resources on APIs in the next 12 months, compared to just 11% planning to reduce investment. More than half of respondents (54%) said better connected services and faster feature delivery create customer value that translates to business value.
Overall, the survey finds 82% of respondents work for organizations that have adopted some level of an API-first approach to building and deploying applications, with 25% operating as fully API-first organizations. A total of 43% of fully API-first organizations generate more than 25% of total revenue from APIs, with 20% of those organizations generating more than 75% of total revenue from APIs. A full 74% generate at least 10% of their total revenue from APIs.
However, the survey also finds most organizations will soon encounter multiple API challenges in the age of AI. In the past 12 months, Postman has already seen 7.53 million calls made to AI APIs, a 40% year-over-year increase. OpenAI dominates AI traffic (56%), racking up 4.2M calls over the past 12 months but Gemini and Llama have both seen 3.1x and 6.9x year-over-year growth, respectively. The survey, however, finds 60% of respondents are still designing APIs primarily for humans only rather than other applications such as artificial intelligence (AI) agents. Less than a quarter (24%) actively design APIs with AI agents in mind, with only 13% designing equally for humans and AI agents.
The imbalance will need to be addressed in the months ahead, with many organizations needing to apply more rigor to any API that will be exposed to AI agents, said Balaji Raghavan, head of engineering for Postman. AI agents, for example, will expect APIs to be well documented, he adds. “The documentation will need to be easily discoverable,” said Raghavan.
Additionally, many organizations will also need to consolidate APIs that have overlapping functionality to ensure AI agents are not confused, he noted.
Developers will also need to become familiar with the Model Context Protocol (MCP), an emerging de facto standard for connecting AI agents and APIs, said Raghavan. While 70% of developers are aware of MCP, the survey finds only 10% are using it regularly.
More than half of respondents (51%) are also worried about unauthorized or excessive API calls from AI agents, followed closely by AI systems accessing sensitive data they shouldn’t see (49%) and AI systems sharing or leaking API credentials (46%).
On the plus side, survey respondents are already using AI to improve code quality (68%) and generate API documentation (41%). Top obstacles for AI adoption are a lack of trust in AI systems (36%) and ethical, legal, and compliance concerns (33%).
Unsurprisingly, REST remains the dominant API framework (93%) but usage of other approaches such as (50%), WebSockets (35%), and GraphQL (33%) continue to grow.
Ultimately, APIs are going to be more crucial than ever in the age of AI. The survey finds 69% of respondents are now spending more than 10 hours per week on API-related tasks. However, 93% also acknowledged teams struggle with API collaboration, leading to duplicated work, delays, and degraded quality. Specifically, inconsistent, outdated, or missing documentation creates confusion about API behavior (55%) while other teams admit to rebuilding functionality that already exists because they can’t discover or access existing APIs (43%).
One way or another, digital CxOs are, in many cases, finally going to need to address these API issues head on if they expect their organizations to stay relevant in the AI era.
