AWS Makes Kiro Bots Generally Available With Persistent Memory and Scheduled Jobs

Quick Facts

• AWS made Kiro bots generally available on November 5, 2026, inside Kiro Crew, its open-source multi-agent development workspace.
• Each bot maintains persistent memory through a feature called Notes, capturing developer preferences, prior decisions, and project constraints across sessions.
• Amazon Q Developer is being retired, with new signups blocked since May 15, 2026, and full end-of-support set for April 30, 2027.

AWS officially launched Kiro bots into general availability on November 5, bringing a new layer of autonomous coordination to its Kiro Crew workspace, according to the announcement. The bots are designed to own specific slices of a developer's workflow rather than execute isolated tasks inside individual agent sessions.

Kiro first launched July 14, 2025, reached general availability on May 7, 2026, and added Kiro Crew on August 4, 2026. The November release makes bots, the central feature of Kiro Crew, production-ready.

What Kiro Bots Do

Each bot takes responsibility for a defined area of work. An intake bot reads new tickets, labels them, flags likely duplicates, and drafts reproduction steps for bugs. A build-health bot checks the CI pipeline every 30 minutes and separates new failures from known flaky tests.

Bots launch multiple worker-agent sessions in parallel, monitor their progress, redirect deviations, and consolidate results. They seek developer approval only when a decision requires human sign-off.

Developers can schedule recurring jobs directly from chat. That means a bot can review overnight CI failures each morning or draft release notes every Friday without manual prompting.

Persistent Memory and Scoped Permissions

The Notes feature gives each bot a running record of what it has learned, including developer preferences, past decisions, requirements, and constraints gathered from prior sessions and feedback. Developers can inspect, edit, or delete any Note to adjust bot behavior.

Permissions are scoped by role. A research bot can search the web and read codebase files but cannot edit them, while a deployment bot can execute scripts but requires explicit user sign-off before making changes. What one bot learns stays isolated from every other bot's memory.

Model Choice and Infrastructure

Coordinating bots and their worker sessions can run on different models. Kiro's catalog includes models from OpenAI, Anthropic, and open-weight alternatives. Bots run on a personal laptop, a dedicated desktop machine, or an Amazon EC2 instance. AWS is also adding a Dashboards feature for visualizing bot activity across sessions.

Pricing

The Free plan includes 50 credits. Pro costs $20 per month with 1,000 credits. For teams of at least two people, AWS offers pooled usage at $20 per seat per month, with each seat contributing 500 credits to a shared pool.

Amazon Q Developer Retirement

The Kiro launch coincides with the formal wind-down of Amazon Q Developer. New signups have been blocked since May 15, 2026. IDE plugins and paid subscriptions reach full end-of-support on April 30, 2027. AWS has stated that moving a user from Q Developer to Kiro is a one-way migration with no path back to a Q Developer Pro subscription.

Analyst Concerns

The coordination model carries real cost risks. Dave McCarthy, Group VP at IDC, warns of a "coordination tax" where parallel background retries and tool calls rapidly compound token consumption and compute costs. CIOs evaluating adoption should account for that overhead.

Auditability is a second concern. As bots evolve through weeks of interactions and feedback, tracing why a specific decision was made across multiple sessions becomes harder, even with Notes and work logs available for review.

Stephanie Walter, Practice Lead at HyperFrame Research, described the product's value plainly: Kiro Bots attempts to take over more of the coordination burden, resulting in less supervision of individual sessions. More than 100,000 developers joined the Kiro waitlist after its initial launch before AWS removed the waitlist entirely.