AWS Kills The AI Services It Launched Just Two Years Ago
AWS recently moved approximately 20 services, including Amazon Kendra, Q Business, and Bedrock Agents, into maintenance mode, halting new customer sign-ups. This significant pruning is unusual as many retired AI services were launched very recently, some in late 2023. Maintenance mode allows existing users to continue but stops new feature development. This move reflects AWS's strategy to consolidate its AI offerings, replacing point solutions with broader, agent-centric successors like Bedrock Knowledge Bases, Amazon Quick Suite, and Bedrock AgentCore. The aim is to streamline its AI portfolio into three core anchors, mirroring consolidation efforts by Microsoft and Google. While this offers clearer boundaries, it places a migration burden on early adopters and could impact buyer confidence in the longevity of AWS AI services. Enterprises are advised to prioritize abstraction and align with these new foundational services.
AWS announced a service availability update on June 30 that moved roughly 20 services and features into maintenance mode. The list includes Amazon Kendra, Amazon Q Business and Amazon Bedrock Agents, which has been renamed Amazon Bedrock Agents Classic. These services stop accepting new customers a month later.
Cloud providers prune their catalogs all the time, and AWS has been trimming since 2024, when it discontinued AWS App Mesh and a handful of other services. What sets this wave apart is the age of the retirees. AWS shipped Bedrock Agents in November 2023 and Q Business less than a year later. The company is now placing AI services into maintenance faster than many enterprises complete a single procurement and deployment cycle for the same products.
The June announcement goes well beyond the three headline names. AWS reported that 10 Amazon SageMaker AI features moved to maintenance in the same update, including Ground Truth, Clarify, Debugger and Model Monitor. Simple AD and Amazon Cognito Sync joined them. A March update had already pushed AWS App Runner, AWS CloudTrail Lake and AWS Audit Manager into maintenance. That earlier wave also sent Amazon WorkMail and Amazon RDS Custom for Oracle into sunset. Taken together, the two announcements represent the broadest coordinated pruning the AWS catalog has seen.
Why would AWS retire AI services at the peak of enterprise AI spending? What does the company expect its customers to run instead? The answer to both questions is consolidation, and the successor map makes the strategy legible. Every retired AI service has a designated successor that is younger, broader and more agent-centric than the product it replaces.
Kendra customers are pointed to Amazon Bedrock Knowledge Bases, a managed retrieval-augmented generation service with built-in connectors, hybrid search, and an agentic retrieval API. Q Business gives way to Amazon Quick Suite, the platform AWS introduced in October 2025 by folding Q Business capabilities into Amazon QuickSight. Bedrock Agents Classic hands over to Bedrock AgentCore. AWS now positions AgentCore as its foundation for running production agents.
With this consolidation, AWS is collapsing a sprawl of point AI services into three anchors. Bedrock carries the models and the retrieval layer, AgentCore carries agent execution, and Quick Suite carries the business user experience. Enterprises can standardize on fewer services with clearer boundaries, and AWS can concentrate its engineering investment instead of spreading it across overlapping products.
Microsoft and Google Cloud arrived at similar destinations earlier. Microsoft folded its enterprise AI assistants under the Copilot brand, and Google Cloud consolidated its offerings under Gemini Enterprise. The key difference lies in the path, where the rivals built one flagship assistant from the start. AWS shipped Kendra, Q Business and Bedrock Agents as separate products, and it is now unwinding that portfolio in public.
The cost falls first on the customers who trusted the first generation. An enterprise that standardized on Kendra for retrieval two years ago now faces a second migration to Bedrock Knowledge Bases. AWS's own migration guide carries a dedicated section on feature gaps and workarounds. Some Kendra data source connectors lack a native equivalent in the successor, and AWS recommends routing unsupported sources through Amazon S3.
The Q Business path has its own friction. The migration guidance steers customers toward Model Context Protocol integrations for connectors that Quick Suite does not natively support. Those integrations cannot serve as knowledge base data sources for document indexing. None of the June announcements carries a hard end date for existing workloads, which softens the immediate pressure but leaves planning horizons open.
The deeper cost shows up in buyer confidence rather than in migration hours. Enterprise buyers evaluate cloud services on longevity. A platform that retires AI products within three years of launch teaches its customers to discount the next launch.
The takeaway for a decision maker is to treat first-party cloud AI services as a portfolio under active rotation rather than durable infrastructure. The first question for enterprise buyers is abstraction. Which parts of an AI application depend on a specific AWS service API? How much of the logic can move behind an internal interface that survives a successor migration? The second question concerns the durability of the anchors. Does the service sit on Bedrock, AgentCore or Quick Suite, or does it overlap with one of them?
If AWS holds this consolidated architecture through the next two re:Invent cycles, the June pruning will read as a calculated risk that traded short-term migration pain for a coherent platform. Customers who align their roadmaps with the three anchors now will carry less migration debt. Those who wait for the next availability update will let AWS make the decision for them.