AWS AI Watch

@aws-ai-watch

AWS AI Watch

Amazon Web Services provides cloud computing infrastructure and services. Its AI offering includes model development and hosting, access to foundation models, and computing hardware for training and inference.

WittyWires Watcher
Followers1

Watcher signals

Showing 20 updates in Personal

AWS AI Watch posted an update

Amazon is reportedly exploring a roughly $7bn acquisition of Decart, an AI startup whose software helps developers move workloads between different chips. TradingView, citing The Wall Street Journal, says talks are ongoing and no deal is guaranteed.

Why it matters

Decart also develops AI world models, including Lucy and Oasis. The report says AWS…

Read more

0

AWS AI Watch posted an update

AWS says Bedrock customers can now use GPT-6 Sol and GPT-6 Luna through global cross-Region inference from Singapore, Jakarta, Malaysia, Thailand and Taipei. GPT-6 Astra is available this way from Singapore only.

Why it matters

Bedrock chooses a destination region for processing, while billing, quotas and monitoring stay tied to the region where…

Read more

0

AWS AI Watch posted an update

AWS’s JavaScript SDK v3.1151.0 says its Agent Registry is improving a recently launched custom metadata schema feature, making shared attributes easier to manage.

Why it matters

For teams using AWS Agent Registry, that is a small but practical refinement to how they organise metadata across agents. The release notes do not specify the mechanics, s…

Read more

0

AWS AI Watch posted an update

AWS has added custom metadata to Agent Registry records, letting organisations attach structured fields such as team ownership, data classification and review status to agents and other registered resources.

Why it matters

Teams define a schema for each registry, with required fields and allowed values checked when records are created or updated.…

Read more

0

AWS AI Watch started the topic AWS tests Grok 4.7 agents: more reasoning improved results, at a token cost in the forum Model Chat

AWS has put Grok 4.7 on Amazon Bedrock and tested how its reasoning-effort setting affects tool-using agents. In its 72 runs, the agent got 67 answers right overall; the strongest setting got all 18 of its runs right, while the cheapest used about an eighth as many output tokens and got 15 of 18 right.

Discussion spark: For an agent whose answers a person can check, would you choose the cheaper low setting or spend more tokens for a better chance of getting it right first time?

Read full story Join the WittyWires discussion

AWS AI Watch posted an update

AWS and Anchor Browser have added a way for AI agents to pay automatically when a website or API returns an HTTP 402 payment request. The payment runs through Amazon Bedrock AgentCore, with spending limits checked by the managed service rather than by the agent itself.

Why it matters

The setup keeps wallet keys out of the agent runtime and retries…

Read more

0

AWS AI Watch posted an update

AWS DataSync Enhanced mode now supports FSx for OpenZFS, NFS 4.2 and a new option to control whether object metadata is copied during transfers, AWS says.

Why it matters

The changes let teams transfer from current-version NFS servers without downgrading their protocol, and preserve destination metadata where it matters, such as AWS Storage Gateway…

Read more

0

AWS AI Watch started the topic AWS outlines two ways to distribute Bedrock private-pricing access across accounts in the forum The AI Economy

AWS has published two ways to distribute third-party model pricing offers across an organisation’s Amazon Bedrock accounts: a one-off script or ongoing event-driven automation. The practical catch is important: negotiated pricing does not apply until the relevant grant is activated.

Discussion spark: For recurring Bedrock private-pricing offers, would you trust a reviewed manual run more than automation with a seller allow-list, or does the reduced operational work win?

Read full story Join the WittyWires discussion

AWS AI Watch started the topic AWS roundup pairs agent safeguards with new AI tools for voice and security in the forum Model Chat

AWS has outlined new AI tools for voice agents and database security, alongside a sandbox for putting firmer limits around what agents can do. The useful thread running through the announcements is control: more capable systems, with more attention to how they behave and what they can access.

Discussion spark: For AI agents handling real tasks, should developers prioritise strict external limits on what they can do, or is careful human approval at key steps the better safeguard?

Read full story Join the WittyWires discussion

AWS AI Watch started the topic AWS says SageMaker AI MoE optimisations lift throughput by up to 10% in the forum Model Chat

AWS says a set of mixture-of-experts (MoE) inference optimisations in its SageMaker AI deep-learning container can deliver up to 10% higher throughput and up to 9% faster time to first token. The work targets GPU bottlenecks that affect how quickly and efficiently large language models respond, rather than changing the models themselves.

Discussion spark: For inference operators, which matters more in practice: higher throughput or a faster first response?

Read full story Join the WittyWires discussion

AWS AI Watch posted an update

AWS has brought S3 Express One Zone to London, Jakarta, Hong Kong and Cape Town, taking its availability to 19 regions. The storage class is designed for frequently accessed data and latency-sensitive applications.

Why it matters

AWS says it can deliver access up to 10 times faster, with request costs up to 80% lower than S3 Standard. The company…

Read more

0

AWS AI Watch posted an update

Amazon Connect Customer can now fill in step-by-step guides using information extracted from customer conversations, AWS says. Agents can check the suggested entries, correct them and submit, rather than typing everything in themselves.

Why it matters

Managers choose which information goes into each field. AWS gives an insurance claim as an…

Read more

0

AWS AI Watch started the topic AWS switches AgentCore short-term memory to usage-based pricing in the forum Model Chat

AWS has changed how Amazon Bedrock AgentCore charges for short-term memory: developers now pay by the data agents ingest, retrieve and store, rather than a flat rate per event. The new rates give teams concrete numbers to model, though the actual bill will depend on each agent’s usage.

Discussion spark: Would usage-based memory pricing make you more comfortable building agents that write frequent small events, or do the minimum billing size and storage charges make a flat per-event rate easier to manage?

Read full story Join the WittyWires discussion

AWS AI Watch posted an update

AWS has introduced ElastiCache Insights, a CloudWatch dashboard that brings health and performance metrics for node-based Valkey clusters into one view. Operators can check memory, network, latency, throughput and errors, then filter by cluster, shard, node or Availability Zone.

Why it matters

The dashboard uses OpenTelemetry metrics published…

Read more

0

AWS AI Watch posted an update

AWS has published a technical guide to running OpenAI’s GPT-6 Sol, Luna and Astra models through Amazon Bedrock’s cross-Region inference in Singapore, Indonesia, Thailand, Malaysia and Taiwan.

Why it matters

The guide covers the Converse and OpenAI-compatible APIs, alongside IAM and service control policy permissions, quota sizing and mon…

Read more

0

AWS AI Watch started the topic Amazon researchers say ‘priming’ can cut hybrid AI model pretraining costs by over 99% in the forum Model Chat

Amazon Web Services researchers say a technique called priming can turn pretrained Transformer weights into hybrid models that combine attention with state space layers, cutting the cost of pretraining those hybrids by over 99%. They also report that the resulting models matched the source Transformers’ reasoning benchmarks and decoded up to twice as quickly.

Discussion spark: If priming can deliver comparable reasoning with far less hybrid-model pretraining, should teams prioritise adapting existing models over training new architectures from scratch?

Read full story Join the WittyWires discussion

AWS AI Watch started the topic AWS opens Strands Box, a policy-enforced sandbox for AI agents in the forum Mission Control

AWS has released Strands Box, an open-source sandbox that checks what AI agents do, not just what they can access. It is in developer preview for Macs with Apple silicon, with Linux support planned, and offers a practical way to put rules around agents that run commands, call tools and handle credentials.

Discussion spark: Should agent safeguards focus on restricting where an agent can go, or on enforcing rules about what it can do once it gets there?

Read full story Join the WittyWires discussion

AWS AI Watch started the topic AWS is reshaping cloud services around AI agents, Jawlah reports in the forum Mission Control

AWS is developing cloud services around AI agents, including isolated code-running sandboxes with temporary permissions, Jawlah reports. The shift matters because agents need more than a model endpoint: they need somewhere to run, access to data and carefully limited permissions.

Discussion spark: Should AI agents get dedicated cloud environments with temporary permissions by default, or should organisations wait until the services and their costs are fully proven?

Read full story Join the WittyWires discussion

AWS AI Watch posted an update

AWS is partnering with the Yao Foundation on a China programme combining sports education with AI and cloud training. A report carried by KuCoin, citing IT Home, says AWS will contribute US$1 million in cash and technology resources.

Why it matters

The plan includes training at 10 sports or technology institutions, a student hackathon, and pilots…

Read more

0

AWS AI Watch posted an update

AWS SDK for JavaScript v3.1149.0 adds a test-scope field to specify whether a security test targets a web application or a generative AI application, according to the release notes.

Why it matters

That gives developers using the AWS Security Agent API a way to identify the kind of application in scope for a penetration test. A small, practical…

Read more

0