AI.info
AI Tools Updates — Page 172
What changed in the AI tools this week — dated, from the makers’ own changelogs.
In inglese
- scite — June 2026 Release Notes
Manage Collections inside the MCP, find your next question, and set up teams on Scite.
- Coveo for Workplace — Conversational search using the Coveo Search Agent
A conversational search experience using the Coveo Search Agent is now generally available.
- SkillCorner — RC Deportivo Partner With SkillCorner
- Zilliz Cloud — 🔒 Zilliz Cloud Bring Your Own Cloud Infrastructure (BYOC-I) now supports Google Cloud Platform (GCP) .
BYOC-I supports Google Cloud Platform (GCP).
- Altana Atlas — Week of July 6, 2026
Classification results are folded into one check; Entry Processing is now Current, Transshipment is Misdeclared Origin, and entries export to CSV.
- SkillCorner — SkillCorner Releases New Peak Velocity Metric
Releases New Peak Velocity Metric.
- Altana Atlas — Classification, misclassification, and vague-description results are consolidated into a single Classification check; the entry-list tab and the misdeclared-origin flag get clearer names; and you can
Classification, misclassification, and vague-description results are consolidated into one check; you can now export an entry to CSV.
- Unstructured — 0.24.0
The v2 HTML parser now derives category_depth from heading level; URL fetching uses host validation and default timeouts.
- Peec AI — A new Query Fanouts page, plus AI Shopping in the MCP
Query Fanouts shows searches run by AI engines; AI Shopping is now in the MCP and Customer API.
- Bland — Contact Memory Sync to HubSpot & Salesforce
Memory syncs to HubSpot or Salesforce after every call and message, and can pull CRM context into contact memory.
- Sana Learn — Bulk course conversion
Convert PDFs or SCORM packages into courses in one click. Edit mode turns each file into a course.
- Auphonic — Studio Voice Beta Model Released
Studio Voice Beta Model released.
- Sana — Bulk course conversion
Convert multiple files into Sana-native courses in one click.
- Meticulous — 📊 Detect FE Performance Regressions & Custom Checks (Beta)
Custom checks detect regressions in React-component re-renders, network requests, JS bytes loaded, and JS statement execution count.
- Meticulous — 🤖 Agent Skills
Agent skills let your agent pull test run results or find sessions impacted by local changes and run them.
- Meticulous — 💬 Comment on Diffs (Beta)
You can drop a comment on a diff and have the conversation right inside Meticulous.
- Zilliz Cloud — 🔒 Zilliz Cloud Bring Your Own Cloud Infrastructure (BYOC-I) now supports Google Cloud Platform (GCP) . For details, refer to Deploy BYOC-I on GCP for step-by-step manual guides, and Terraform Provide
BYOC-I now supports Google Cloud Platform (GCP).
- InvokeAI — InvokeAI v6.13.6
Corrects a bug that caused Qwen Image and Qwen Image Edit models to crash with a dictionary key error.
- LocalAI — v4.6.0
ROCm backends run on GPU; chat supports conversation forking, and PII/audit events have a Prometheus counter.
- InvokeAI — InvokeAI v6.13.5
Adds support for PEFT named-adapter LoRAs and a Qwen Image single-file checkpoint loader with fp8 support.
- Ludwig — v0.17.7
Fixed Ray preprocessing tests for Arrow-backed data and row ordering.
- Core42 — Advanced Usage Dashboard
Enhanced the Advanced Usage Report with a detailed view for selected usage metrics and an API request breakdown by API key.
- Renderforest AI Logo Maker — Meet Text Layer in Video Editor
Add and style text directly inside Video Editor; adjust size, color, stroke, and alignment.
- Emergent — Introducing the Emergent MCP Connector for Claude and ChatGPT
The Emergent MCP Connector for Claude and ChatGPT
- Frase — Engineers stopped prompting their AI. Content teams should too.
Engineers stopped prompting their AI and started building loops.
- Emergent — Introducing the Emergent MCP Connector: Build Apps Straight From Claude or ChatGPT
Build Apps Straight From Claude or ChatGPT.
- Kagi Assistant — Heads, tails, and an AI toggle
Assistant bug fixes covered font size, code display, MP3 summarization, stopping responses, exports, and model settings.
- Instantly — Go from idea to automation in seconds
Describe what you want in plain language and Instantly Copilot will help create the workflow.
- Kagi Search — Heads, tails, and an AI toggle
Search adds controls to turn off AI features and disable widgets; dice can roll any number of sides, and the widget can flip coins.
- Poolside Assistant — Introducing Laguna XS 2.1
Today we're releasing Laguna XS 2.1, a version of our Laguna XS.2 model.
- Ninja AI — Fable 5 Is Now Live in Ninja Max Mode: What It Is, What It Costs Elsewhere, and Why That Matters
Claude Fable 5 powers Max mode in Ninja.
- Retell AI — Conductor, higher free-tier token limits, and more
Conductor is live; every change gets simulated before it ships; free LLM token allowance raised to 4,000; documents parse more accurately.
- Mistral AI — Leanstral 1.5: Proof Abundance for All
Leanstral 1.5: Proof Abundance for All.
- Synthflow AI Voice Agent Builder — Claude Code & Codex Connectors, Zoom & RingCX, Cross-Channel Memory, Custom Caller ID on Warm Transfers & more
Claude Code and Codex connectors, Zoom and RingCX integrations, shared memory across voice, SMS, and WhatsApp, and caller-ID controls on warm transfers.
- ClickUp Brain — Brain²: Release Notes Takeover
Brain² can generate slide decks, dashboards, landing pages, apps, and prototypes using workspace context.
- Nabla Copilot — New Server API endpoints for E/M (Evaluation & Management) CPT code generation (U.S. product release only):
E/M CPT code endpoints; ICD-10 normalization engine; async normalized data endpoints; ASYNC_CORRECTION; note customization; ChartSync source_note.
- Harness CI — July 2026
Added global queuing for macOS builds on Harness Cloud.
- Granola — Granola for Android is here!
Android users can take notes anywhere with their phone.
- PLaMo Translate — Proプラン・Proチームプランの提供を開始しました
テキスト翻訳やファイル翻訳の利用上限を引き上げた「Proプラン」「Proチームプラン」を提供開始しました。
- Browser Use Cloud — Browser Use CLI 3.0
CLI 3.0 allows agents to execute Python freely within the browser, so agents can do more actions and can inspect, adapt, and recover.