AI.info
AI Tools Updates — Page 79
What changed in the AI tools this week — dated, from the makers’ own changelogs.
- Replit — Work across projects from a chat
You can now ask Replit to find your existing projects, read their files, and send tasks to them, all from a chat.
- Replit — Create and manage TikTok Ads
Create, launch, and track TikTok ads from a chat in Replit.
- Replit — Replit Desktop Windows preview with Microsoft and NVIDIA
Replit Desktop can now build and run apps locally on your Windows PC, available in private preview.
- Jan — Jan v0.8.6: AppImage, Turing GPU, MTP Speed & MCP Image Fixes
Fixes Linux AppImage model-load crashes, CUDA engine support on NVIDIA Turing GPUs, Qwen3.8 MTP speed, and MCP tool images reaching the model.
- Botpress — Features & enhancements
Added support for GPT-6.1 Sol, Claude Sonnet 5.5, Claude Haiku 5.5, and GLM 5.3 Flash.
- Claude Code — 2.1.296
Added gateway code policy settings, subagent compaction and model controls, and an overload retry delay setting.
- Intercom Fin — Macro Admin Mode
Teammates with the Can manage shared macros permission can now view everyone’s personal macros in Macro settings.
- Cerebras Inference — Clarified service-tier fallback and response behavior
Responses report the effective tier in service_tier; auto can fall back to flex and priority can retry at the request’s standard priority.
- Canny — Follow ideas
Admins can follow ideas and get notified when they receive a new insight or their status changes.
- vLLM — proto-v0.5.0
Released vllm-proto 0.5.0.
- Gram — Dashboards get a shared, savable filter bar, and Explore adds a skills dataset and distinct-value counts
Dashboards gain shared savable filters; Explore adds a skills dataset and distinct-value counts, and its query engine to Platform MCP.
- Profound — SEO data is now in Profound, analyze dashboards with your AI Marketer, and more
AI Marketer can analyze Moz SEO data for every page, and Pages now includes a Search Visibility tab.
- OpenHands Cloud — openhands: v0.80.0
OpenHands, Agent Canvas, and automation image tags were bumped; Laminar, Valkey, and Keycloak fixes were included.
- Pixverse — PixVerse Growth Studio Integrates with Meta’s Catalog Product Video, Helping Merchants Extend Video Across Their Full Product Catalog
Integrates with Meta’s Catalog Product Video, helping merchants extend video across their product catalog.
- Sigma — Assistant in build mode: new capabilities
Assistant in build mode supports dynamic layouts, page duplication, background images, waterfall charts, progress bars and rings, and layout settings.
- CrewAI — 1.15.27
Added XPU for OpenCLIP, deepinfra provider support, evaluation stop reasons, and per-run cost and time tracking.
- QA Wolf — v1.43.0
agent get --sessions --follow follows multiple agent sessions and returns the first completion, failure, cancellation or question.
- Appsmith — Release v2.5.0 🌈
Appsmith AI reached end of life: new connections and queries can’t be added, and existing queries have stopped working.
- Neon — All AI Gateway models with prepaid credits, backend features with your coding agent, and new ways to send feedback release - Oct 09, 2026
AI Gateway models became available with prepaid credits, and Neon added backend features for coding agents and new feedback options.
- Heuritech Market Insights — TikTok Fashion Trend Data Is Now LIVE on Heuritech
TikTok is now part of Heuritech's fashion trend data, alongside Instagram and Little Red Book.
- Make — Make AI Agent improvements, new AI models, SSO login routing, and MCP tools for audit logs
Make AI Agent fallback connections and prompt caching, Maia, Claude Sonnet 5.5, GPT-6.1 Sol, SSO login routing, and MCP tools for audit logs.
- ThoughtSpot Analytics — Enhancement
You can now access memory generation for Liveboards from the Spotter optimization tab on a Model.
- Snowflake Cortex Analyst — Agentic Marketplace Discovery (Public preview)
Agentic Marketplace Discovery
- Fern SDK Generator — 3.105.1
reference.md is now generated for every output mode.
- Fern SDK Generator — 3.105.0
Support inferred bearer auth with type: refresh-token; the SDK always sends grant_type: "refresh_token" to the get-token endpoint.
- Ollama — v0.40.2
Models downloaded with earlier Ollama versions are upgraded in the background on first run, with originals kept as backups.
- Tonic Textual — Textual now saves the dataset sort order between user sessions.
- Raycast — ⚡ More performance improvements and new ways to send context to AI
AI Chat, Notes, and Settings now run in separate processes; Send Element to AI lets you send screen content to AI Chat.
- Tembo — Agent Instruction History, Browser Notifications, and Claude Haiku 5.5
Agents keep a history of saved instructions; browser notifications alert when an agent needs your approval; Claude Haiku 5.5 is in the model selector.
- Birdeye Reviews — Weekly Product Release Notes | October 9, 2026
Catch up on the latest capabilities, feature enhancements, bug fixes, and platform improvements across Birdeye.
- Qdrant — v1.19.0
Adds TurboQuant 4-bit primary vector storage, per-component memory settings, prefix keyword filtering, and per-query IDF corpus for sparse search.
- Letta — Release v0.34.8
Browser discovery now detects Letta Code on the device.
- Mistral Vibe — Managed Deployments
Managed Deployments let you run Workflows without provisioning or operating your own worker infrastructure.
- GLM Coding Plan — Release v3.14.5
After a manual reset, the number of remaining reset chances is now retained in the display.
- Snowflake Data Cloud — Agentic Marketplace Discovery (Public preview)
Agentic Marketplace Discovery
- Dataiku LLM Mesh — LLM Mesh
- RunPod Pods — New Release Global volumes for Serverless (Beta)
You can now attach a global volume to GPU Serverless endpoints from the console.
- XO — RC-20 Retro Color v1
Updated platform; Added notification feature; Many bugfixes and UI improvements
- Retell AI — Introducing Conversation Memory
Agents can now remember a caller across conversations, with a short brief for each contact and settings to save, recall, view, edit or clear briefs.
- Suki Platform — Ask: seamless information retrieval and synthesis at the point of care
Seamless information retrieval and synthesis at the point of care