AWS is also rethinking its pricing plan for the agentic AI-driven IDE.
AWS is limiting usage of Kiro, its agentic AI-driven integrated development environment (IDE), and introducing a waitlist for new users, just a week after announcing its public preview.
It has also deleted details of the pricing tiers it planned for the service, saying “We’re reviewing our approach to better align with how developers are using and want to use Kiro.”
“To ensure a smooth and consistent experience for everyone, we’ve introduced some temporary measures — including a waitlist for new users and daily usage limits for existing users,” Jay Raval, a specialist solutions architect at AWS, told Kiro users.
Users were initially told they would get up to 50 free agentic interactions with the service each month, but the exact number they will get remains a mystery for now.
“While we’re not sharing specific limits just yet, that’s only because things are still evolving,” Raval wrote on Kiro’s official Discord channel. “We’re actively monitoring system performance and adjusting as needed based on real-world usage patterns.”
AWS initially said it would offer three tiers of service for Kiro: free with a cap of 50 agentic interactions per month; Pro at $19 per month for up to 1,000 interactions, and Pro+ at $39 per month for up to 3,000 interactions.
But initial feedback from developers prompted AWS to review its pricing plans for Kiro, with developers concerned about how the service counts different interactions and tasks completed, and how it accounts for specifications when using different models, the company wrote on the pricing page on the Kiro website. This now says that, while Kiro remains free within “reasonable limits” during the preview period, “updated pricing details for different tiers will be shared soon,” and encourages users to “continue providing feedback.”
Users seek predictable pricing
The changes have not gone down well with developers, with some taking to social media platform Reddit to call for the usage limits’ removal or to vent their frustration about how they are “killing the workflow.”
Others, tempted away from Anysphere’s rival agentic IDE, Cursor, after price changes there found themselves facing renewed price uncertainty with Kiro.
Forrester vice president and principal analyst Charlie Dai said, “Rising costs for advanced AI models and infrastructure scaling challenges are forcing platforms like Cursor and Kiro to adjust pricing and impose waitlists and limits.”
But such changes are making it hard for users to evaluate the costs and benefits of using such platforms.
That, said Dai, risks alienating individual users. “Poor communication usually erodes trust and drives users toward transparent competitors. Trust hinges on pricing clarity and reliability”, he said.
Developers’ rising interest in Kiro can be attributed to its spec-driven development workflow. This is typically used by large engineering teams to plan, design, and ship code, and differs from that adopted by other AI coding assistants which rely on ad-hoc prompts.
In the context of Kiro or any other agentic IDEs, spec-driven development follows a process wherein a developer or a team of developers first defines what the application should do, and then an agent or agents build and test it to match the specification defined by the team.
Latent problems
Early users of Kiro have reported that the platform’s responses have been slow, to the point where tasks don’t get completed correctly.
Kiro uses Anthropic’s Claude Sonnet 4.0 and 3.7, and such reports about long latencies in response times for Claude models are not new, with the company’s status page showing 21 incidents around model issues already in July. In June, there were at least 23 incidents and in May, the model provider faced over 35 incidents.
The incident count for Claude models in July is a strong indicator of underlying infrastructure strains, said Steven Dickens, CEO and principal analyst at HyperFRAME Research.
Users of Kiro have also pointed out other flaws with the agentic IDE with its official GitHub page showing 804 requests to resolve issues at the time of filing.
The page also shows that at least 367 issues were resolved forcing AWS to launch an update of the platform — version 0.1.15 — that AWS’ Raval said will provide improved prompt caching, improved login reliability, better error-messages, faster model switching, and refined trusted commands.
