Daily Tech Update

The Enterprise Site Management APIs now use a separate API key permission scope for each API, replacing the single Manage Site SSL & Publish State scope: Manage Enterprise Site Publish State for the Site Publish State API . Manage...

The Stores Ribbons V3 API now includes a Count Ribbons method. Use it to count the ribbons that match a filter, or to get the total number of ribbons on the store if you don't pass a filter. Count Ribbons supports the same filters as Query...

The Blog Draft Posts API has 2 new methods for managing draft posts in multiple languages: Translate Draft : Creates a draft post in another language by copying an existing draft or published post. The new post shares the source's...

The Remittance Invoices API lets you retrieve the daily records of how much of a merchant's sales goes towards repaying a Wix Capital cash advance. Each invoice covers a single sales day and splits the amount into the portion applied to...

The Site Management APIs include: Site Publish State : Take a published site offline without deleting it, so it stops serving visitors while it stays in the account and remains editable. SSL Certificate : Install your own SSL certificate...

The Async Site Translator API translates a site's content in bulk, in the background. Start a job with Translate Site , then poll Get Translation Job Status to follow the job's progress and retrieve any fields it couldn't translate. Scope...

The Premium Assets V2 API lets you retrieve what a Wix account pays for. A Premium asset is a read-only view of a purchased Premium product, together with the product it came from.

OpenAI released GPT-6.1 Sol, which it says approaches its top Astra model on coding and computer use at one-fifth of Astra's API token prices, making it a cheaper option for coding workloads.

OpenAI's DevDay covered more than 20 announcements across GPT-6 Astra, Codex, the APIs and new builder tools, which is the single place to review what changed for API integrations.

Claude Opus 5, Sonnet 5 and Haiku 4.5 are now available on Amazon Bedrock with in-country inference in India, so data can stay within the India Regions.

OpenAI introduced Dots, always-on background agents that act across connected apps and learn user preferences, a new agent surface that apps may need to integrate with.

OpenAI added features that let ChatGPT discover and run third-party software for both people and AI agents, creating an alternative distribution channel to app stores.

Bedrock now offers Claude Opus 5 and Sonnet 5 with in-region inference in Seoul and Claude Sonnet 5 in Singapore, helping meet local data processing requirements.

An architecture guide argues for using plain deterministic code wherever a responsibility can be precisely specified and reserving LLMs for judgment and ambiguous input, which helps decide where to put model calls.

A write-up on source-aware verification for MCP agents describes checking where a tool-provided fact came from and not just whether it is correct, which is useful when building agent pipelines.

OpenAI launched a suite of office-style features inside ChatGPT, extending it into document and productivity work that clients may start to compare with existing tools.

Meta is opening its Muse AI agent to small businesses for running operations and finding customers, which may come up in client conversations.

NVIDIA published Kumo Tabular, a model for tabular prediction that it reports improves the accuracy and efficiency tradeoff for structured data tasks.

Part two of an Amazon Quick series covers prompt patterns and pitfalls for Quick Research, Flows, Sight, chat agents and action integrations.

A case study shows how Condé Nast built multimodal video search over more than 140,000 videos on Amazon Bedrock, cutting a task that averaged 250 minutes, which is a reference design for media search.

An AWS reference build uses Amazon Quick and Bedrock AgentCore agents to extract and verify contract fields and answer portfolio-wide questions where plain RAG chat falls short.

Herb, an HTML-aware ERB parser merged into Rails core on August 25, is slated as Rails 8.2's opt-in view engine and underpins ReActionView, which adds reactivity to plain ERB templates.

Vercel Connect now accepts third-party service submissions to its connectors directory, with OAuth or API key connection methods, so services no longer wait for Vercel to build integrations.

Amazon RDS now reports full snapshot size through a new FullSnapshotSizeInBytes field in DescribeDBSnapshots and a console column, making incremental snapshot storage costs easier to check.

Apache Polaris 1.8.0 changes stale-commit responses from HTTP 400 to 409 and unknown commit states to 500, so clients that handle those codes need review before upgrading.

DumboDB 0.7.0 adds Merge Modes and fixes a bug where MongoDB compare-and-swap version increments used for optimistic locking silently failed on its Dolt-backed versioned storage.

A hands-on report on Xcode's MCP Preview tool for SwiftUI lists pitfalls such as no choice of render device and stale screenshots, so AI-driven preview workflows need workarounds.