AI.info
AI Tools Updates — Page 127
What changed in the AI tools this week — dated, from the makers’ own changelogs.
En anglais
- Speakeasy SDK Generation — The gateway refuses AI agents captured by Shadow AI that your organization has blocked, and Pi sessions land in the dashboard under policy
The gateway enforces Shadow AI blocks for agent callers, and Pi sessions and policy decisions appear in the dashboard.
- Speakeasy SDK Generation — Claude Code and Codex tokens and cost reappear on usage, cost, and identity pages
Native OTLP usage now appears on usage, cost, and identity pages.
- Typefully — 1080p Video Publishing on X
X videos published through Typefully now support 1080p when the account has an active X Premium subscription.
- Fern Docs — Site-level changelog for multi-product sites
Add a top-level changelog key in docs.yml and the changelog is served at /changelog.
- Fern Docs — Reference API specs directly from docs.yml
Point an api navigation item at OpenAPI, AsyncAPI, or GraphQL specification files with a specs list, without creating a generators.yml.
- Elicit — Turn your Elicit Library into a shared research hub
Research Agent can now work from your Library, teams can maintain collections together, and Projects can prioritize the evidence that matters
- Gemini Enterprise Agent Platform — Gemini Omni Flash supports stateful and streaming video generation (Preview)
Supports stateful and server-sent event (SSE) streaming video generation in the Interactions API; generated videos and interaction state can be stored.
- Ask Lyra — Introducing Lyra Agent: The AI Agent for Customer Feedback Analysis
Introducing Lyra Agent for customer feedback analysis.
- Ask Sage — v3.7.3
Single sign‑on now counts as multi‑factor for API‑key (programmatic) access
- Ask Sage — v3.8.0
Superadmin Admin Activity view; token‑distribution accuracy, government model failover & Section 508 accessibility hardening
- Zed — Zed 1.20.2
Fixed provider payment errors showing a Zed Pro upgrade prompt instead of the provider's original error.
- Devin — September 16, 2026
Voice calls use a live speech model, sessions wake as you type, and copied messages preserve formatting.
- Hightouch — [Improvement] Customer Studio
Audience size limits accept a percentage of members and apply after filters.
- Cody by Sourcegraph — Sourcegraph self-hosted 8.0.0 is now available
- Splice Sounds — Splice x Spotify Fresh Finds Forward: Helping emerging artists “Make It ‘Till You Make It”
Eligible artists in Spotify’s Fresh Finds Forward program can get six months of Splice’s Creator plan.
- UiPath Maestro — Conversational agents in Maestro Flow (Preview)
Maestro Flow can now model AI-powered chat and voice agents, with intent-based routing and hand-off between specialist agents.
- UiPath Maestro — Native document processing nodes (Preview)
Document processing nodes bring UiPath document intelligence onto the Flow canvas as discrete steps.
- UiPath Maestro — Maestro Flow is now generally available
Maestro Flow is now generally available. It is the third canvas in Maestro Orchestrate, alongside BPMN and Case.
- UiPath Maestro — Inline data entity modeling (Preview)
In Maestro Flow, you can now create and edit Data Fabric entities directly inside a VS Code solution.
- UiPath Maestro — Autopilot for Flow (Preview)
Autopilot scaffolds a flow from a description of intent, so you start from a working structure instead of an empty canvas.
- UiPath Maestro — Coded Functions as nodes (Preview)
A coded function can now be called from the Flow canvas as a node.
- Lokalise — Retrieve AI Scoring results via the API
Retrieve existing AI Scoring results alongside translations through the Lokalise API using include_ai_scoring=1.
- Hightouch AI Decisioning — [Improvement] Customer Studio
Audience size limits can use a percentage of members and apply after filters.
- Claude Code — 2.1.275
Added syncing of claude.ai skills and plugins to terminal sessions, with separate opt-outs.
- Claude Code — 2.1.274
Added Claude apps gateway store.connect_timeout_seconds to configure the Postgres connection timeout.
- Intercom Fin — See who's viewing a conversation, now in Table view
Teammates' avatars appear on conversation rows in inbox Table view to show when a colleague is already handling a conversation.
- Devin Desktop — September 16, 2026
Devin calls use a live speech model; sleeping sessions wake as you type; copied messages preserve formatting.
- Pika — Welcome to the New Pika
The new Pika brings video, image, and sound creation into one place, with gen media models underneath and creative apps on top.
- vLLM — proto-v0.3.0
Released vllm-proto 0.3.0.
- vLLM — v0.30.0rc1: [Bugfix] Isolate supplemental FlashInfer BF16 autotuning (#57285)
Isolated supplemental FlashInfer BF16 autotuning.
- Retool Agents — Embed apps in iframes
Apps can now be embedded in other apps and in external apps or websites using an iframe.
- Gram — The gateway refuses AI agents captured by Shadow AI that your organization has blocked, and Pi sessions land in the dashboard under policy
The MCP gateway blocks agents blocked in Shadow AI, and Pi sessions and policy decisions appear in the dashboard.
- Gram — Claude Code and Codex tokens and cost reappear on usage, cost, and identity pages
OTLP usage appears across usage, cost, and identity pages, with fixes to the usage timeline and log detail sheet.
- Skyvern — Release v1.0.53
Gated passkey 2FA behind the enterprise plan and added a book-a-demo upsell tile.
- beehiiv — More Connected Communities
New tools for data, discovery, and engagement.
- Unsloth — Docker + Multi User + AMD Support
Adds a Docker image with NVIDIA and AMD support, multi-user accounts, diffusion support, ARM64 Windows CUDA support, and RDNA1/2 support.
- Riverside.fm — AI Image & Video Generation
Generate b-roll, guest cards, transitions, and more, with AI models, right in Riverside.
- Retool — Embed apps in iframes
Apps can now be embedded in other apps, classic apps, and websites using an iframe.
- Keploy — v3.6.65
- Keploy — v3.6.66
Auto-replay no longer auto-passes additive response-schema changes.