AI agents, from demos to real work.
Useful autonomy / Clear boundaries / Human control
AWS recaps September additions to Bedrock, AgentCore and Strands
AWS’s September roundup combines a public-preview managed-agent service, serverless runtime changes, broader model availability and automated knowledge-base synchronization for enterprise builders.
3 min read
Postman documents how Agent Mode limits tools and context while scaling on Amazon Bedrock
The vendor-published implementation guide says production reliability depended less on adding agent capabilities than on constraining tool exposure, shaping context and matching Bedrock routing and caching…
5 min read
AWS details pay-per-inference payments for AI agents with BlockRun and Incarna
AWS says its AgentCore payments service let Incarna deploy budget-governed x402 payments for individual BlockRun inference calls, with customer-owned wallets and per-request settlement.
4 min read
Microsoft Research open-sources Agent Lightning v1.0 for RL training with deployed agent harnesses
The lightweight framework uses an LLM proxy and Kubernetes-native rollout execution to connect existing agent harnesses to reinforcement-learning workflows without reimplementing their control loops.
4 min read
Cornerstone says Orion AI cut database diagnosis from 45 minutes to 10
An AWS customer case study details how Cornerstone combined domain-scoped agents, live-data access and confirmation gates for database operations, while reporting a 78% reduction in diagnosis…
4 min read
AWS adds SageMaker inference optimization skill for coding agents
The aws-ai-ml skill produces reviewable SageMaker SDK v3 code for live endpoint benchmarks, deployment recommendations and benchmark comparisons, with safeguards around traffic-generating tests.
4 min read
AWS documents agentic retrieval for multi-part RAG queries in Bedrock Knowledge Bases
AWS’s guide frames planning-based retrieval as a higher-cost, higher-latency option for multi-part questions that a single search is less likely to cover completely.
5 min read
AWS documents a three-layer framework for evaluating explainable multi-agent systems
The reference implementation separates response quality, business-rule validation and explainability so teams can pinpoint whether an agent’s problem is its decision-making or its explanation.
4 min read