Official announcements and practical decisions
AI Tool Change Watch
Catch up on official AI tool announcements, understand what to recheck, and follow updates for your saved tools.
- 61
- recorded announcements
- 54
- tools covered so far
- 2026-10-15
- announcement sources checked
AI Tool Change Watch
What changed. What to check.
Selected official announcements, checked 2026-10-15. 54 tools covered so far; other services are not yet tracked here.
61 updates
No updates match this view. Clear the filters or save a tool from the directory.
Plan a migration for Pipedream Workflows before March 2027
Pipedream schedules Workflows and String shutdown for March 31, 2027, with Workflows data deletion by April 30. Connect is unaffected. The notice does not state its publication date; October 15 is our verification date, not the shutdown or announcement date.
Check before use Inventory dependencies and export each project. Check the export manifest and preserve needed data separately; JSON exports are rebuild references, not automatic migration or credential backups. Test replacement triggers and reconnect accounts before cutover. Do not start a new long-term Workflows deployment.
Prepare important Gems for the transition to skills
Google's guide dates the Gemini app skills rollout to October 13. Gems remain usable during transition. Workspace business/enterprise retirement is no earlier than March 1, 2027; education no earlier than June 1. These are not fixed deadlines for every account.
Check before use Inventory important instructions and Ask a Gem flows. Test replacements with non-sensitive examples and check administrator settings. Skills do not currently sync between Gemini and Workspace; plan any manual recreation. Migrated draft skills require review and activation.
Use LM Studio for Bionic's local inference
LM Studio 0.4.26 lists coexistence with Bionic: while both apps run, Bionic can use LM Studio for local model inference. The note requires Bionic 1.1.8 or newer.
Check before use Check both installed versions and the selected model endpoint. Test a small task and monitor memory use; local inference alone does not mean connected tools or every part of an agent workflow stay offline.
Distinguish Ollama's new cloud connector from local use
Ollama announced a connector for using its cloud models inside Fly.io Sprites. This is a cloud execution option, not an update that makes those models run on your laptop.
Check before use Before connecting, check both services' costs, data policies, and which Sprites have access. Start with non-sensitive input and minimum permissions; do not assume a cloud sandbox guarantees confidentiality.
Review agent edits and database access in Notion
Notion 3.8 announces clearer page and database change history with attribution, plus database column permissions for people and agents.
Check before use Before connecting a team agent, test access with a low-privilege account and inspect its edits. Check your plan and retention options; visible history is not a substitute for a backup or permission audit.
Recheck the voice model before revising an existing project
Descript's October release notes introduce ElevenLabs v4 for AI Speech and warn that some projects will switch models when reopened. The model can be selected in AI model settings.
Check before use Keep an approved audio reference before regenerating lines. Compare voice identity, pronunciation, pacing, and edit boundaries against it. Use only authorized voices and confirm speaker consent, current access, and usage costs before delivery.
Check legacy HeyGen plan rights before changing a subscription
HeyGen's October 11 guide update warns that leaving a legacy Unlimited plan is permanent. Eligible existing subscriptions can continue while active and in good standing, but canceling or switching to a credit-based plan does not allow a return. This is plan guidance, not service closure.
Check before use Identify your current plan before changing it for more credits. Review whether a confirmation affects only an add-on or the main subscription. Ask support about unclear controls rather than canceling experimentally or buying a second subscription to force a change.
Check the Jan hotfix before replacing a model that stopped loading
Jan 0.8.6 addresses regressions after 0.8.5, including Linux AppImage model-load crashes on some distributions and CUDA loading failures on older NVIDIA Turing setups. The release also fixes image results from MCP tools reaching vision-capable models.
Check before use Record your app version, operating system, GPU and error before troubleshooting. Back up important settings and test a small known model after updating from the official release. Confirm model compatibility and tool permissions; do not assume the fix guarantees accuracy or makes remote providers private.
Recheck execution permissions before letting an agent run commands
Cursor IDE 3.24 adds approval checks for certain risky delete commands in Auto-review. Its release notes also say Run Everything runs agents without the sandbox unless permissions.json restricts that mode. These are different execution modes, not a blanket safety guarantee.
Check before use Check your installed version, execution mode and permissions.json before continuing repository work. Keep a recoverable copy of changes, protect secrets, and avoid unrestricted execution in untrusted projects. Review commands, dependencies and generated diffs, then run tests before merging.
Work across existing projects from one chat
Replit says a chat can find projects, read their files, and send background tasks for review. Its announcement says these changes are not published automatically.
Check before use Confirm the source and destination projects before sending work. Review permissions, copied code, and each diff; test the destination app before publishing.
Refine clip timing and fades before exporting a mix
Suno's Studio update adds drag-based time stretching and adjustable fade curves, alongside changes to downbeat detection. The release describes Studio access for Premier subscribers, not every account.
Check before use Check current plan access and export terms first. Test a short passage, listen for stretched-vocal artifacts and abrupt fades, and verify timing in the exported mix. Use samples you have permission to use; editing features do not establish commercial-use rights.
Review Claude's policy update before November 12
Anthropic announced a revised Usage Policy taking effect November 12, 2026. It clarifies existing high-risk-use requirements and adds requirements for autonomous physical actions.
Check before use Teams using Claude for consequential decisions should read the full policy and confirm qualified human oversight and disclosure. This is a forthcoming vendor-policy update, not legal advice or a new permission to automate decisions.
Check which brand defaults Jasper applies outside its app
Jasper's September roundup, published October 7, says MCP generations and agent runs use workspace Style Guide, Brand Voice and Audience defaults when IQ is not explicitly selected. Explicit selections take precedence. API integrations can opt into a workspace-default fallback; this is not automatic for every integration.
Check before use Have the workspace owner review defaults before using connected workflows. Compare a small test with implicit defaults and explicit IQ selections, and inspect the API option separately. Check source material, permissions and factual claims before sharing; inherited brand settings are not a compliance or accuracy guarantee.
Describe scenario versions before changing an automation
Make's updated toolbar adds descriptions to saved scenario versions and displays them in version history, helping users choose a version to restore.
Check before use Save a named baseline before editing. Test with sample data and inspect activation settings before resuming; restoring a scenario does not undo emails already sent or records already changed.
Check local agent sandbox permissions
GitHub announced local sandboxing for Copilot CLI, the Copilot app, and VS Code sessions using Agent Host. Policies can restrict file, network, and credential access.
Check before use Check support and active policy in your client before use. Test with a disposable repository and limit permissions; sandboxing does not replace code review or protect every secret automatically.
Give the design agent your library rules before generating changes
Figma's October 6 agent release adds configurable library guidelines and file search in chat as the agent exits beta. Markdown guidelines are read when prompting with that library enabled. The announcement also says the agent starts using AI credits on October 6.
Check before use Try a small approved library with explicit component and brand rules. Confirm the library is enabled, inspect referenced files and generated edits, and check current credit usage and access before wider adoption. Guidelines do not guarantee correct or accessible designs.
Try an interactive block before your next presentation
Pitch introduced the Agent-powered Anything block for prompted slide elements such as calculators and tabbed views, using the presentation theme's colors and fonts.
Check before use Prototype with non-sensitive sample data. Verify calculations, keyboard access, and behavior in the exact sharing or export format your audience will use; do not assume interactions survive a static export. Check current access and usage terms.
Prioritize workflow fixes before an n8n upgrade
n8n 2.43 release notes describe clearer migration-report impact categories and a progress summary of workflows compatible with the target version.
Check before use Check availability on your instance, back up configuration, and test affected workflows in staging. Start with upgrade blockers and execution failures; a compatible label is not a substitute for testing real inputs.
Follow local agents from the Cursor iOS app
Cursor announced remote viewing and replies for local agents through its iOS app. The agent stays on your computer, which must remain on and online.
Check before use Review device pairing and workplace policy before enabling access. Cursor says remote control defaults on outside Enterprise organizations; check your settings and keep reviewing changes before release.
Replace Gen2 streaming dependencies in Murf voice workflows
Murf's October 5 changelog confirms that Gen2 is decommissioned for the Streaming API and points users to Falcon 2. Gen2 remains available for non-streaming text-to-speech. This is a streaming-model retirement, not closure of Murf AI.
Check before use Check which model and endpoint an existing voice integration uses. For streaming, test a Falcon 2 migration with non-sensitive scripts before production. Review pronunciation, timing, errors and current costs rather than assuming an identical replacement; keep non-streaming workflows distinct.
Check whether a citation supports your claim
Consensus introduced Skills in Deep mode, including Verify Citations: a workflow that compares a draft's claims with cited sources and shows passages behind its assessment.
Check before use Start with one non-sensitive section. Read the original paper and check its methods, population, and context before accepting a verdict. Follow academic disclosure rules; an AI citation check is not proof of accuracy.
Explore file data with charts inside Computer
Perplexity's October release describes interactive visualizations directly in Computer threads, including turning a CSV file into an interactive chart.
Check before use Try a small, non-sensitive file first. Check totals, units, missing values, and source dates against the original data before using a chart in a report; confirm access and usage costs in your account.
Test a video idea in Draft mode before finishing
Runway's October 2 release adds Seedance 2.5 Draft mode in Agent, with lower-resolution drafts and post-generation enhancement.
Check before use Use a short draft to check composition and motion before finishing. Verify credit costs and output settings, then inspect product details, text, likenesses, and media rights again in the final video.
Recheck your video workflow budget after Synthesia's plan update
Synthesia's notice sets October 1 as the effective date for self-serve plan changes, including renaming Creator to Pro and revising included allowances and credit usage. The notice cautions that API or bulk-personalization workloads may stretch credits less far. This date is the effective date, not an inferred publication date.
Check before use Inspect your account's plan, remaining credits and billing details rather than relying on old screenshots. Estimate one representative video, dubbing or bulk job with current in-app costs before scaling up. Confirm applicable terms and access with support if your account differs from the notice.
Keep an evidence review current with Routines
Elicit introduced Routines for recurring Research Agent work, including finding new evidence and preparing an updated artifact for review. Instructions and schedules can be edited before activation.
Check before use Define the research question and inclusion criteria first. Inspect original papers and citations before changing conclusions, check current usage terms, and keep confidential material out unless your policy permits it. Monitoring does not guarantee complete coverage.
Review dictated text before it becomes a message or work instruction
Fireflies announced Talk in its desktop app for Mac and Windows: hold a key to dictate, then release to insert text at the cursor. The announcement describes cleanup and formatting for the destination app. This September release is newly tracked here.
Check before use Check current app support, permissions and data policies before enabling dictation. Try non-sensitive text in a draft field first. Verify names, mentions, dates and commitments before sending, and do not dictate passwords or confidential material without workplace approval.
Check access before connecting your Canva brand assets to an agent
Canva announced a connector for Meta's Muse to create editable designs using account and brand context. The announcement specifies US and Canada availability; it is not a worldwide access promise. This September update is newly tracked here.
Check before use Confirm regional and account availability, then inspect requested permissions before connecting. Start with non-confidential brand assets. Check text, image rights and brand consistency in Canva before exporting or sharing; do not assume an agent-produced design is approved for publication.
Test delivery instructions on a short voiceover first
ElevenLabs introduced Eleven v4 with natural-language delivery instructions and inline audio tags for directing speech. The announcement covers tone and pacing controls; it is not a guarantee of consistent performance for every voice or language.
Check before use Compare a short tagged take with an untagged reference before generating a full script. Check names, numbers, emphasis, and unintended sounds. Obtain voice permissions, protect private recordings, and follow applicable synthetic-audio disclosure rules before sharing.
Update Speechify new-voice onboarding for verified speaker consent
Speechify's September 23 API notice replaces the old consent field with a speaker challenge and recorded response across API versions. Existing cloned voices and synthesis are unaffected. This changes new voice creation, not the availability of the entire Speechify service.
Check before use Review any cloning integration and test the official consent flow with a consenting speaker. Do not rely on an old API version to bypass verification. Contact support for migration issues, and check voice rights and intended use separately from the vendor's verification step.
Evaluate Next Gen Zaps without replacing live workflows
Zapier announced early access to Next Gen Zaps, including branches that rejoin and runs that pause for a human reply. Access is requested rather than assumed available to every account.
Check before use Confirm eligibility and billing before trying a small workflow. Test approval pauses, error paths, and duplicate handling with non-production accounts before migrating a working Zap.
A missing Open WebUI Integrations tab does not mean connections stopped
Open WebUI v0.11.4 makes the personal Integrations tab opt-in through an administrator's Direct Integrations setting. The release notes explicitly say that hiding the tab leaves existing connections working. This is a control-visibility change, not removal of the integration service.
Check before use Ask your instance administrator to review existing tool and terminal connections separately from the tab's visibility. Do not treat the hidden tab as a revocation mechanism or enable access indiscriminately. Back up and test a supported release before upgrading a shared instance.
Check model replacements before October 19
GitHub's September announcement schedules selected Copilot models for retirement on October 19, 2026. The notice lists affected models and suggested replacements.
Check before use Compare your model choices with the official list and test a supported replacement on a familiar task. Check organization model policies; this is a scheduled change, not a claim that retirement has already happened.
Check which Tabnine CLI your scripts run before migrating
Tabnine's September 15 release introduces its OpenCode-based CLI. Separate legacy CLI documentation says active development stopped as of 6.6, with critical updates through December 31, 2026 and deprecation afterward. This record uses the release date, not an inferred notice date. Tabnine itself is not shutting down.
Check before use Read the legacy CLI notice linked on the tool page. The new installer reassigns tabnine to the OpenCode CLI and renames the old command tabnine-legacy. Audit scripts, settings and permissions with your administrator before migration. Use current supported releases; the release notes mark 6.6.0 unstable, so this is not a recommendation to install it.
Connect source summaries with quizzes and flashcards
Google's Gemini Notebook announcement describes a rollout of interactive learning overviews combining summaries with study outputs such as quizzes and flashcards. Check availability in your account rather than assuming the rollout is complete.
Check before use Use material you are allowed to upload, try answering without notes, and check generated answers against the original sources. Correct misleading questions before sharing, follow academic rules, and avoid uploading sensitive documents without approval.
Turn a repeated team task into a shared skill
Notion 3.7 introduced a shared library of reusable agent instructions for tasks such as document review and business briefs. This September announcement is newly tracked here.
Check before use Start with one task, an approved example, and a named owner. Test outputs and sharing permissions, remove sensitive examples, and recheck exported copies when the original skill changes.
Check old COPILOT formulas before recalculating an Excel workbook
Microsoft retired the preview COPILOT worksheet function on September 14. Previously calculated values can remain visible, but recalculation returns #NAME?. Copilot in Excel's pane continues separately; this is not retirement of all Copilot features.
Check before use Work on a backup and locate COPILOT formulas before refreshing a report. Verify important cached results against their sources. Review a replacement workflow through the Copilot pane if available for your account, rather than assuming it is a drop-in formula replacement.
Review comments can close after a follow-up commit
GitHub reports that Copilot can resolve its own addressed review comments during a rereview, alongside changes to review analysis.
Check before use Inspect the actual diff and test results before merging. A resolved AI comment is not evidence that a fix is correct.
Desktop MCP clients can connect to Replit
Replit's release note describes connecting compatible desktop MCP clients to work with apps, including publishing actions.
Check before use Review the connected account, exposed actions, and approval settings. Start with a disposable app before allowing changes to production work.
Recheck folder access after Beautiful.ai's library migration
Beautiful.ai's September 10 update replaces separate personal and team folder structures with private or shared folders and says existing folders migrate automatically. Folder sharing and new User Groups are scoped to Team plans; groups support presentation and folder sharing, not comments, Assign Slide or nested groups.
Check before use Inspect who can open a migrated folder and its presentations before adding client material. Check guest and group membership with a workspace owner, then test access using the intended recipient account. Do not assume an old folder name proves it is private or that every collaboration feature supports groups.
Projects introduces coordinated, longer-running work
Cursor announced a beta rollout of Projects, with a coordinator delegating work to agents and maintaining shared project context.
Check before use Check rollout access, execution locations, connected services, and how to stop work before testing. Keep secrets out of shared context and review each resulting change.
Check existing Windsurf conversations before moving to Devin Local
Windsurf's official changelog, now Devin Desktop, says version 3.9.19 removed Cascade and provides a Continue in Devin Local prompt for existing conversations. This September transition is newly tracked here, not an October launch.
Check before use Back up unfinished work before migration. Check the selected repository, branch or worktree, execution location and permissions; do not assume conversation migration transfers uncommitted files. Verify workplace policy and secret access, and review and test changes before merging.
Investigate the factors behind a changing business metric
Zoho Analytics On-Premise build 6440 added Key Driver Analysis in Ask Zia for Linux environments. This is a specific Analytics deployment update, not a claim about every Zoho product or account.
Check before use Confirm your edition and build, then inspect a known metric with a small dataset. Check filters, time windows, missing values, and source totals. Treat suggested drivers as hypotheses, not proof of causation, before making business decisions.
Test multilingual meeting notes on the device you actually use
Granola's September announcement expands desktop multi-language mode, with Korean among the listed languages. Users select Multi-language in language settings. The notice explicitly separates desktop coverage from mobile and watch support; this is a historical update newly tracked here.
Check before use Check support for your device and meeting languages, obtain required participant consent, and test a short non-sensitive conversation. Verify names, numbers, decisions and action owners before sharing notes. Language support does not guarantee transcription or summary accuracy.
Relay.app has closed; check workflows that depended on it
Relay.app's official notice says it shut down in September 2026 and permanently deleted all accounts, workflows and run history. No exact closure day is supplied. This is a closure notice, not an instruction to export data from the discontinued service.
Check before use Review dependencies and connected-service permissions. Use only information or exports already retained independently; do not assume login, export or recovery remains available. Test replacement workflows before enabling real business actions.
Hand off tables that recipients can still edit
Gamma's August release notes describe editable tables in PowerPoint exports and in Google Slides after upload, with an opt-out when visual fidelity matters more. This is an earlier release newly tracked here.
Check before use Export a sample deck and open it in the recipient's app. Check table values, wrapping, fonts, and editability before delivery; retain a reviewed PDF reference and confirm permissions for client data and images.
Review Fathom recording and sharing defaults separately
Fathom's August 20 admin guide explains separate auto-capture and visibility defaults for internal, external and unscheduled meetings. Team settings can override organization defaults. Turning auto-capture off does not prevent manual capture, and changes to auto-capture affect future meetings.
Check before use With your admin, inspect the effective team settings and any visibility locks before sensitive calls. Test a non-sensitive meeting and its audience. Do not treat automatic-capture controls as a blanket recording ban or assume a settings change stops a meeting already being recorded; follow participant-notice and workplace requirements.
Replace shared Copilot podcast links in learning materials
Microsoft's dedicated FAQ confirms Copilot Podcasts retired on August 18. Earlier recordings and shared links are no longer accessible, and those recordings cannot now be exported from Copilot. This concerns Podcasts, not the entire assistant.
Check before use Audit reading lists, training pages and handoffs for Copilot podcast links. Use retained source material or independently saved copies where permitted, and review any replacement summary. Do not promise that an old recording can still be downloaded or recovered.
Choose reminder or recording behavior separately for each meeting app
Otter's August desktop update introduced meeting detection with Off, Remind me to record, and Record automatically options per meeting type. Its announcement identifies reminders as the default, not automatic recording. This earlier release is newly tracked here.
Check before use Review Meeting Detection settings in your desktop app and check current availability. Obtain required participant consent and follow workplace recording policy before enabling capture. Test reminders first, verify stop controls and sharing permissions, and disable capture for sensitive conversations.
Turn support analysis into a reviewed improvement proposal
Fin's official FAQ dates Operator's general availability to August 6 as part of the Pro add-on. Its current documentation describes support analysis, content-gap investigation, and proposed guidance changes for team review.
Check before use Confirm permissions, usage costs, and analysis limits. Compare findings with representative conversations rather than assuming complete coverage. Review proposed changes and test responses before enabling them; keep customer data within approved access boundaries.
Review automations that depended on Tapo or Kasa through IFTTT
TP-Link's US support notice sets August 1, 2026 as the end of IFTTT services for Tapo and Kasa products. This is the effective date, not the notice publication date. The affected device integrations ended; this does not mean the IFTTT platform closed.
Check before use Identify affected Applets and test device behavior rather than assuming a routine still runs. TP-Link points to Smart Action in its apps; verify your device, region and required triggers before rebuilding. Check for duplicate actions and keep manual control available. Do not assume a replacement reproduces every IFTTT workflow.
Remove assumptions about author metadata in You.com API results
You.com's July 15 changelog marks removal of the authors field from Web Search API v1 results and Answer API source results as a breaking change. Its legacy v0 search API was unaffected. This concerns API response metadata, not a shutdown of You.com or a change to every consumer search feature.
Check before use Check parsers, citation templates and stored-result migrations for a required authors field. Test a current response and handle missing authors explicitly. Verify attribution against the original source; do not fabricate an author or silently treat a domain name as one. Review the current schema before deploying an integration change.
Keep a human review step in a connected translation workflow
DeepL's July 7 Translation Flow article describes applying glossaries, style rules and translation memories, then inviting language experts to adjust translations. This record uses the dated article, not a claim that Translation Flow first launched in July or is newly released in October.
Check before use Confirm availability, costs and connector permissions for your account. Test a non-sensitive sample with an approved glossary; compare meaning, obligations, dates, numbers and layout with the original. Require a qualified reviewer for sensitive material before publishing translated content.
Replace Motion 1.0 dependencies in Leonardo video workflows
Leonardo's June 4 notice sets June 17, 2026 at 09:30 AEST (UTC+10) as the end of Motion 1.0 availability across its API, web app and mobile apps. The retired model is identified as svd. This is a model retirement, not closure of Leonardo AI.
Check before use Check saved video workflows and API model settings for svd before rerunning an old project. Leonardo points to Hailuo 2.3 as a replacement; verify current access, inputs, output behavior and cost rather than assuming identical results. Test a short non-sensitive sample and review rights before client delivery.
Review generated test cases before enabling a store workflow
Shopify's May Flow update lets Sidekick find existing shop data for test events covering workflow paths. Merchants can review, edit, remove, or add cases. This earlier announcement is newly tracked here, not an October launch.
Check before use Include normal orders and exception cases, not only the case that prompted the workflow. Check whether a test run can trigger real actions before executing it, use a safe test setup, and protect customer data. Generated cases do not guarantee complete coverage.
Guide image edits with markup and variation controls
Previously announced in April: Adobe introduced AI Markup for visually targeted edits and Precision Flow in beta for exploring image variations. This is newly tracked here, not a new October launch.
Check before use Try an owned image and compare edited areas with the original. Check current access and beta status, inspect unintended changes to logos or products, and confirm rights before client delivery.
Trial AI batch edits on a few products before exporting a catalog
Photoroom's March product roundup added AI tools to web Batch workflows and image renaming before export. AI batch options had plan-dependent access. This is a historical release newly tracked here, not an October launch or a claim that mobile has identical features.
Check before use Check current platform, plan and usage limits. Test a small set of approved product photos, compare colors, labels, shape and accessories with originals, and verify SKU filenames after export. Keep original files and reject edits that imply features the product does not have.
Check grammar in Korean and other added languages
Grammarly's March announcement added beta grammar and spelling support for languages including Korean through its browser extension and desktop app. The announcement treated advanced tone and clarity suggestions as future work, not part of that release.
Check before use Check current support in your app with a short, non-sensitive draft. Review names, numbers, politeness, and intended commitments yourself; a grammar correction does not verify facts or guarantee appropriate tone. This March release is newly tracked, not an October launch.
Adjust speech, music and background noise separately in Adobe Podcast
Adobe's March 5 update describes Premium-only Enhance Speech source separation: separate controls for speech, background noise and music, plus downloads of isolated tracks. This gives editors a more selective cleanup workflow than relying on one processed mix.
Check before use Check current plan access, retain the original recording and test a short clip. Listen to the separated tracks and final mix for missing speech or unwanted artifacts. Removing music does not itself establish copyright clearance; check recording consent and media rights before publishing.
Use current Midjourney support channels instead of old Rooms instructions
Midjourney's February 26 announcement removes Rooms from its website. The notice directs users needing help to its help page, Discord support channels or support email. It describes future work without giving a replacement launch date. This does not announce closure of Midjourney.
Check before use Update team onboarding links and old tutorials that send users to Rooms. Check current official channels before arranging a live session or collaboration workflow. Do not assume an old room link still provides support, or that a promised replacement has launched; avoid posting private client material in public help channels.
Confirm Udio delivery options before planning a downloadable soundtrack
Udio's official UMG partnership help page states that audio, video and stem downloads are disabled. February 17, 2026 is the date displayed on that help page, not a verified start date for the restriction. This is a historical restriction rechecked today, not a new shutdown announcement.
Check before use Do not promise downloadable files or stems to a client based on older tutorials or a subscription alone. Verify current delivery options and applicable rights before starting paid work. A shareable Udio link is not a substitute for an export when the project needs local files; use only material you can lawfully deliver.
Audit documents and integrations before moving a legacy Frase account
Frase's new-platform announcement is dated February 2 and updated July 21. Its current migration guide, linked on our tool page, says there is no required switch deadline. It warns that only live documents migrate and that folders, integrations and Content Governance need setup again. These are current migration checks, not a newly dated shutdown announcement.
Check before use Have the account admin inventory active and archived work, retain backups and notify teammates before migrating. Check plan eligibility and usage allowances in your account. Rebuild required settings and verify access and content before reconnecting publishing workflows; do not assume every old setting carries over.
Announcement checks are separate from full tool reviews. Availability and terms may change. Unlisted changes are unknown, not proof that a service is unchanged. Read status stays in this browser.
Start with a task
All guides
Aiplorer review log
Aiplorer records when public tool pages were checked against official sources and when practical Guides and Use Cases were reviewed for scope, safeguards, and current workflow fit. The announcement watch above records selected vendor changes and practical follow-up checks. The review log below records Aiplorer’s separate editorial checks; a new review date does not itself establish a product change.
Use the Review Activity Finder to focus the chronological log by content type or tool category. If this browser has a local shortlist, the finder can narrow the log to saved candidates. On a return visit, the private browser checkpoint can also surface entries added or checked again since the previous snapshot.
AI services and recommended workflows can change between reviews. Recheck official pricing, availability, privacy, licensing, limits, and product documentation, and review AI-assisted output before making an important decision.
Return visit checkpoint
See the latest Aiplorer review activity
The latest recorded review is 2026-10-15. This browser can remember the current public review set and identify later tool, Guide, or Use Case activity when you return.
Stored only in this browser. Review activity does not mean that a vendor changed its product.Review activity finder
Focus on the checks that matter now
Choose a content type or tool category to scan the public review log. This changes the view, not the chronological order.
92 review checks in this view
Reset the view or choose another content type or tool category.
Aiplorer review activity
Aiplorer review activity
Aiplorer review activity
Aiplorer review activity
Aiplorer review activity
Aiplorer review activity
Aiplorer review activity