Why they built it
Jean Yang saw software becoming a collection of services whose interactions were difficult to understand, with the API as both a source of complexity and a useful place to observe it. Akita first built an API fuzzer that sent varied inputs to find defects, betting teams would pay to test APIs for security problems.
How it ended
Akita did not shut down as a failure; Postman acquired it in July 2023 and brought its technology and team into its API platform. Postman said it acquired Akita to add discovery and monitoring alongside its existing tools for designing and testing APIs. Financial terms were not disclosed in the announcement.
What went wrong
Most important first.
- 1
Fuzzer needed API specs customers lacked Product-market fit
Teams liked the fuzzing demonstrations but lacked specifications describing their APIs, and Akita spent months producing a specification before testing a prospective first major customer's system in 2019.
- 2
Pivot to observability took a year and different skills Team
Yang said moving from fuzzing to API observability took roughly a year and required different skills.
- 3
Standalone discovery lacked established workflow distribution Distribution
Postman already held collections of API requests used in development while Akita added a view of interfaces actually receiving production traffic, and Postman said it acquired Akita to add discovery and monitoring.
- 4
Faced entrenched API-management and observability incumbents Competition
Akita competed for attention and budget with observability and API-management products including SwaggerHub, MuleSoft, Azure API Management and Google Apigee, where a gateway can manage traffic at an existing control point while Akita sought to infer what was happening across services.
What could be different today
- Postman now offers Insights as an Enterprise feature that infers endpoints from production traffic and reports errors and latency, so discovery is bundled with its existing design and testing workflow.
- Postman's Insights documentation lists deployment and traffic limitations, supporting specified REST environments and noting lack of support for encrypted traffic and for gRPC and GraphQL in the listed troubleshooting guidance.
- API gateways and existing monitoring tools already give teams partial views of their interfaces, so a rebuild must prove a different deployment or governance advantage rather than assume discovery is unavailable.