AI.info
AI Tools Updates — Page 132
What changed in the AI tools this week — dated, from the makers’ own changelogs.
- Runway — Agent in Asset Comments
You can now @-mention Agent directly in comments on any asset page.
- Notion AI — Notion 3.7: Agent skills for your whole team
Notion 3.7 adds agent skills for your whole team.
- Claude AI — Launching Salesforce in Claude (beta)
We've launched Salesforce in Claude, a plugin that brings a seller’s accounts, opportunities, and pipeline into Claude with 37 pre-built sales skills.
- Spotter — Fixed an issue where the indexing service experienced a rapid surge in memory usage when processing high-cardinality columns with long text or URL values (>2000 bytes).
The indexing service now handles high-cardinality data without unbounded memory growth.
- Glide — Apps can trigger workflows, and 18 smaller improvements and fixes.
The agent can connect workflows to your app, so a button can start a workflow.
- LiteLLM — v1.101.0
Heuristic auto router, semantic MCP tool search, off-peak pricing
- AirOps — Social & Community Insights, Prompt Changelog, and Editable Campaign Criteria
Explore the YouTube videos and Reddit discussions that appear in AI answers, and understand changes to the prompts behind your reporting.
- LangGraph — 2026-09-07 - LangSmith Cloud update
Annotation queue feedback follows the resolved reviewer identity, and invalid SDK feedback is recorded on runs instead of dropping the ingest.
- LangChain — 2026-09-07 - LangSmith Cloud update
SDK feedback that violates workspace key configuration now records error feedback on the run instead of rejecting the ingest request.
- Grammarly — Bug fixes and performance optimization
- Unstructured — 0.27.6
DOCX text_as_html now represents merged-cell text with colspan and rowspan instead of duplicating it.
- RunPod Pods — Deprecation Nano Banana Edit
The Nano Banana Edit public endpoint will be retired on September 28, 2026, because Google is discontinuing the underlying model. Migrate to Nano Banana 2 Edit.
- Tonic Fabricate — New setup wizard for self-hosted instances
The setup wizard completes a compatibility check, installs the license key, and provides telemetry options.
- Spotter — Fix
The indexing service now handles high-cardinality data without unbounded memory growth.
- Ramp Bill Pay — Lot-tracked item receipts for NetSuite
Capture lot number, quantity, and expiration date on an item receipt, then sync the full inventory assignment to NetSuite.
- Gumloop — Harrington Harbour
A new Auto mode for all Gumloop agents. It picks the right model for the task, so you don't have to.
- Ramp Corporate Card — Lot-tracked item receipts for NetSuite
Capture lot number, quantity, and expiration date on an item receipt, then sync the full inventory assignment to NetSuite.
- Vapi — Simulations in Composer
Composer can create and run simulations, including scenarios, tester personalities, and suites, so you can test your assistant from the conversation.
- Vapi — Anthropic Bedrock Fallback Models
Set a same-provider fallback model for Anthropic Bedrock in the assistant editor.
- DeepEval — typescript-v0.9.16
New release.
- DeepEval — python-v4.2.3
New release.
- Bolt.new — What is Bolt Forge?
Bolt Forge runs on open-source AI models, giving every individual Pro plan up to 50X more usage at no extra cost through October 14, 2026.
- Raycast — 🌉 macOS Golden Gate Ready
Raycast user interfaces were updated to support macOS Golden Gate.
- Napkin AI — v1.1.26 (2026-09-14)
Requests above the accepted body size and content without a word are refused; failed requests report a reason code where available.
- Meshy — The Image to Image API now accepts input_task_id — chain directly from a completed Text to Image or Image to Image task instead of copying its output URLs into reference_image_urls .
The Image to Image API now accepts input_task_id to chain from a completed Text to Image or Image to Image task.
- Jellypod — Speaker Spacing
Adjust the gap between hosts' takes from one second of overlap to one second of breathing room.
- PromptLayer — Deployment 1
General performance and stability improvements
- Taskade — v6.249.0 — Sturdier Builds & Clearer Automation Runs
App builds resume after connection drops, and automation runs show provider statuses; archived or deleted workspaces stop automations.
- Modal — Upload IdP metadata as XML when configuring SSO
SSO setup now accepts an XML metadata file in addition to a metadata URL.
- RunPod Pods — Nano Banana Edit
The Nano Banana Edit endpoint will be retired September 28, 2026. Migrate to Nano Banana 2 Edit.
- Deepgram — SDK releases
Adds Voice Agent ForceEndTurn support, Flux TTS expressivity controls for Agent providers, and speed range of 0.5 through 1.5 in 0.05 increments.
- LiteLLM — Heuristic auto router, semantic MCP tool search, off-peak pricing
- ELSA Speak — Meeting Coach: Master Every International Meeting with Confidence
Master Every International Meeting with Confidence
- Tonic Fabricate — 4.29.0
After you deploy a self-hosted instance of Fabricate, the first time that you go to the instance URL, Fabricate provides a setup wizard.
- Mixedbread Rerank — Structured Output for Toast 1
The Responses and Chat Completions APIs now support structured outputs for Toast 1.
- Abridge — Pre-Bill Review for CDI and Coding Teams at Partner Health Systems
Pre-Bill Review for CDI and Coding Teams at Partner Health Systems.
- Omni — September 7, 2026
Added undo and redo to the navigation editor, app pagination, map tile controls, and sorting by last login on the Users page.
- Weaviate — v1.37.17 - Modules weaviate_module_error_total metric labels Fix
Adds node-level query admission control and fixes module error metric labels.
- Weaviate — v1.38.15 - Contains filter performance Improevements, Modules weaviate_module_error_total metric labels Fix
Improves Contains filter performance and adds node-level query admission control.
- AirOps — Social insights, prompt history, and editable Campaign criteria
Explore YouTube videos and Reddit discussions in AI answers, review prompt changes, create with GPT-6 Astra and GPT Image 2.5, and refine Campaign criteria.