Changelog

The meaningful changes in Nowledge Mem, written for people using it every day.

Latest release

0.10.67

21. Aug. 2026

0.10 is a ground-up engine rewrite

The 0.10.x line runs on Mem's rebuilt native engine, so most things are faster and lighter. It is a new engine, though, so you may still hit something unexpected. If you do, open Settings → Info, export your app.log, and send it to hello@nowledge-labs.ai and we'll get on it quickly. And if anything is blocking your work, you can download 0.9.29 from the Downloads page and roll back anytime.

0.10.67

21. Aug. 2026

0.10.67Latest release21. Aug. 2026

Safer Answers From Your Own Knowledge

Mem now treats retrieved Memories, Threads, Library items, and Team workspace content as reference material rather than instructions, so agents can use your knowledge without letting old content quietly steer the task.

Highlights

Retrieved knowledge is now handled as evidence, not instructions.
Feed and Knowledge Answer are more resistant to old content steering a new task.

Fixed

Feed and Knowledge Answer agents now keep retrieved content behind a stronger prompt boundary. Memories, Threads, Library artifacts, and Team workspace material are treated as evidence to read, not commands to follow.
This reduces indirect prompt-injection risk while keeping citations, attribution, and Space visibility boundaries intact.

0.10.66

20. Aug. 2026

0.10.6620. Aug. 2026

AI Now Reviews Its Own Skill Edits, and the App Recovers Instead of Going Blank

AI Now can propose skill updates for you to review and reconnects cleanly after a dropped turn, while failed views recover with a retry instead of going blank, and large-library search rebuilds keep getting steadier.

Highlights

AI Now can propose Skill updates for your review, and reattaches instead of dead-ending after a dropped connection.
A failed view now shows a retry instead of taking down the whole window.
Search rebuilds keep getting steadier and more precise on large libraries.

Added

AI Now can propose an update to one of your Skills during a conversation. You see a diff before anything changes, and can approve, edit, or decline it inline.
Settings now shows whether the local CLI is correctly wired to your active remote or Access Anywhere connection, and can repair the wiring in one click when it drifts.

Improved

Editing an AI Identity now happens directly in its reading layout — the name, essence, charter, and voice stay at their normal size and position while you edit, instead of jumping into a form.
Search index rebuilds keep getting more memory-bounded and precise on large libraries: rebuild batches stay within a fixed size, importing new conversations no longer forces a full semantic re-embed, and rare source metadata cases no longer stall a rebuild.
Local context rule files now show the same source, scope, and as-of freshness markers as Nowledge Cloud, closing the last gap in the provenance work started in 0.10.65.

Fixed

AI Now no longer dead-ends into "Use Continue" after a dropped connection while a turn is still running on the server. It reattaches to the live turn instead of asking you to restart it.
A failed request or a rendering error in one view no longer blanks the entire window. Memories, Threads, and every other view now confine the failure to that pane and offer a retry.
Remote MCP OAuth pairing now keeps the authenticated public HTTPS origin ahead of any simultaneously running Quick Link, so a Tailscale Funnel pairing can no longer regress to a rotating trycloudflare.com address.
The desktop app no longer spawns netstat.exe while shutting down on Windows, removing a rare source of shutdown hangs and OS error dialogs.
A provider timeout you configure in Settings is now honored end to end for Background Intelligence and AI Now, instead of being silently overridden by a shorter built-in default.

0.10.65

19. Aug. 2026

0.10.6519. Aug. 2026

A Clearer Home for AI Identity and More Reliable Capture

AI Identities and Team knowledge are easier to understand at a glance, while AI Now sessions, coding-agent conversations, and long Grok chats return to Mem more reliably.

Highlights

AI Now can act as the identity you maintain in Mem, then return its work as a durable Thread.
Session capture survives restarts and conflict recovery without duplicating or dropping turns.
Search rebuilds, backup restores, and repeated PDF imports close several long-running reliability gaps.

Added

In the desktop app, AI Now can run a task as a chosen AI Identity. Completed sessions return to Threads automatically, where they stay searchable and can contribute to future Memories.
Devin and LangGraph now have first-class Mem integrations with scoped Memory access and identity-aware conversation capture.

Improved

Automatic session capture resumes from a durable checkpoint after restarts and sends only new turns. If the destination was deleted or changed, Mem performs one bounded reconciliation instead of duplicating or losing the conversation.
Library Wiki topics keep their names and explanations tied to current Memory evidence, making broad or drifting topics easier to understand and refine.
Team members can review Working Memory rulings and the evidence behind contested claims directly from the Feed, without needing workspace-admin access.
In Nowledge Cloud, scheduled briefings now follow the workspace time zone, including daylight-saving changes.
Nowledge Cloud can schedule deferable Background Intelligence for your preferred local hours, helping you use capacity-sensitive or lower-cost model windows without delaying interactive, manual, or safety work.
AI Identities now have one clear home with a stronger presence, readable profiles, and inline editing. Team brings people, shared knowledge, and usage into three focused views, so the same controls no longer compete across the app.
Nowledge Cloud keeps Threads, Memories, and search results inside the workspace you are viewing. When a personal workspace has no content yet, Mem explains that Team knowledge lives in a separate workspace instead of showing an ambiguous empty screen.

Fixed

Large search-index rebuilds no longer remain trapped in a repeated Message FTS repair loop, especially in long-running Docker libraries.
Backup imports restore Thread and Source content before rebuilding graph links, so restored records no longer point at missing payloads.
Re-uploading the same scanned PDF now recovers its existing Library source instead of failing as a duplicate.
MCP App search cards no longer show a stale graph after the selected result changes, and AI Now artifact previews open reliably.
The browser extension now captures complete Grok conversations opened inside X, including turns outside the visible page, and preserves comparison tables with their rows, columns, links, and code intact. Sync, Distill, and Import restore your scroll position and refuse to save when completeness cannot be verified.
Provider settings now describe model routing as it actually works: most Background Intelligence uses Default, while AI Now, Timeline chat, and other interactive agents use Agents unless you choose Same as default.
The desktop update confirmation now follows your selected app language, including its version details and install or cancel actions.
Connection screens now show the actual workspace address that each client will use, so local, Access Anywhere, and Cloud targets are no longer easy to confuse.
Credit top-up links now open reliably from the desktop app instead of being blocked by the app window.
The desktop sidebar now keeps account sign-out within reach even when the web account menu is unavailable.
The desktop app now verifies that the nmem command actually runs from a new terminal before marking the CLI as installed. When PATH setup is needed, it shows the exact change and asks before applying it.

community-2026-08-18

18. Aug. 2026

community-2026-08-1818. Aug. 2026

Nowledge Mem for Devin

Devin teams can now carry durable context across CLI, Desktop, and enterprise Cloud sessions without mixing account identity with agent memory.

Platform

community-2026-08-17

17. Aug. 2026

community-2026-08-1717. Aug. 2026

Nowledge Mem for LangGraph

A first-class Python connector gives LangGraph agents shared memory without taking over their execution state.

Highlights

One durable memory layer for LangGraph agents, with identity and Thread semantics that survive model and deployment changes.

Platform

LangGraph agents can now load the right Nowledge Agent and Space context on every turn, use scoped memory tools, and keep completed conversations as searchable Mem Threads.
The connector keeps LangGraph checkpoints untouched, treats assistant IDs as deployment metadata rather than conversation identity, and captures a parent/subagent run once instead of duplicating every subgraph.

0.10.64

16. Aug. 2026

0.10.6416. Aug. 2026

Safer PDF Imports and Smoother Setup

Highlights

Searchable PDFs avoid redundant OCR and stay usable when optional vision work fails.
Unsaved Memory edits survive background refreshes and failed saves.
Remote setup, Feed links, and web copy actions are steadier across everyday browser setups.

Fixed

Unsaved Memory edits now survive detail refreshes, failed saves, and changes made while a save is still in flight.
Searchable scanned PDFs now keep their existing text instead of sending every page through OCR. If optional figure analysis fails, the document remains searchable and reports a partial result rather than a failed import.
Remote MCP OAuth setup now keeps the stable public HTTPS hostname that opened the pairing flow, and ChatGPT Subscription no longer offers a Codex-only model that the local runtime cannot execute.
Working Memory links from the Feed now open the exact Memory they point to, including memories that belong to another Space.
Copy buttons in the web and Cloud app now keep working in restricted browsers, local network pages, and other places where the standard clipboard API is unavailable.
Firefox extension setup now keeps the permission prompt attached to your click, so connecting the extension to Mem is more reliable.
Provider setup errors now keep the useful provider detail instead of replacing it with a generic model-capability hint.
The Copilot CLI community plugin now works cleanly on Windows PowerShell.

0.10.63

15. Aug. 2026

0.10.6315. Aug. 2026

Faster Starts and More Reliable Imports

Highlights

Web cold starts download about one-sixth of the previous entry payload.
Long scanned PDFs preserve useful partial OCR results.
Kimi Code imports stay stable across transcript rewrites.

Improved

The web app now paints its loading shell before the API bootstrap completes and downloads only the active language on first load. The initial entry payload drops from about 1.4 MB to 258 KB gzip, so cold starts feel immediate instead of blank.
MiniMax can now be selected as the Vision provider for Library PDF and image parsing. Mem routes vision requests to MiniMax's compatible endpoint even when chat uses its Anthropic-style endpoint.

Fixed

ChatGPT custom MCP connections no longer finish OAuth successfully and then fail with HTTP 422 and zero available actions when the host starts without an MCP session ID.
Long scanned PDFs now keep the pages that fit within the OCR budget instead of failing the entire Library import. Source metadata clearly records partial, skipped, failed, and budget-exhausted work.
Kimi Code session imports now keep stable message identities when transcript files are rewritten, and discovered sessions preserve their Kimi source and project or workspace identity instead of being grouped incorrectly.
Running out of configured graph memory no longer traps Mem in a fatal startup loop. The service stays available in a degraded state so you can raise the limit or export diagnostics, and automatic sizing records a safer floor for the next start.

0.10.62

13. Aug. 2026

0.10.6213. Aug. 2026

DeepSeek Harness, ZCode, and Safer Backups

Highlights

DeepSeek Harness and ZCode are now first-class Mem integrations.
Backups restore Library original files, not just parsed text.
Long remote-embedding rebuilds are bounded before they reach your provider.

Added

Fixed

Cloud restore now brings Library originals back with the backup. PDFs, documents, images, and other uploaded files can be previewed or downloaded after restore instead of leaving only parsed text behind.
Search Index rebuilds no longer send oversized Memory or Library text straight to a remote embedding provider. Long items are bounded before the request, so one large note is much less likely to keep semantic search stuck in text-only mode.
ChatGPT custom MCP connector setup is more reliable when OAuth discovery is probed after Allow. The pairing page now answers the compatibility endpoint without widening the connector's default read-only scope.
Data Transfer and Library recovery now fail more explicitly: recoverable export and restore mismatches become warnings, while binary-looking Library uploads are quarantined instead of being indexed as broken placeholder text.
AI Now no longer automatically replays slow or interrupted model requests. If a local or self-hosted model is still working, Mem now stops and lets you continue deliberately instead of queuing the same large prompt again.
The Threads source filter now treats case variants as the same source. Sessions from Hermes, hermes, and related Hermes agent names appear under one clean source instead of duplicate filters.
Desktop installers and self-hosted Docker images are now built with a more resilient PDF parser dependency download, so brief network interruptions during release packaging are less likely to delay an update.

cloud-2026-08-13

13. Aug. 2026

cloud-2026-08-1313. Aug. 2026

A Better Home for Teams on Nowledge Cloud

A wave of improvements for team workspaces on Nowledge Cloud. They arrive with your workspace, so there is nothing to install or update.

Highlights

Every Mem you belong to is one switch away, with Personal and Team clearly told apart.
Team imports explain who can read what before you commit.

Added

See every Mem you belong to and switch between them. Workspace pickers now tell Personal from Team at a glance, so you always know whose memory you are looking at.
Name your workspace. Rename it right from the switcher, and a new team is invited to pick its name the first time it opens.
Plan and seats in one place. Your team's plan, seat usage, and a direct path to billing now live together on the Team page.

Improved

Signing in takes you to your team first, so shared work is one step closer.
Application and sign-in emails now arrive in your language and address your company by name.
Invites tell the truth. If an invite email could not be delivered, Mem says so clearly, instead of leaving a teammate waiting on a message that never comes.
Importing into a team workspace now explains up front who will be able to read what, before anything is brought in.

0.10.59

13. Aug. 2026

0.10.5913. Aug. 2026

Clearer Setup and Graph Feedback

Highlights

Graph Explorer no longer gets stuck on an empty load.
Distillation tells you when a language model is missing.
Update install stays explicitly user-confirmed.

Fixed

Graph Explorer inside MCP-capable clients no longer sits on an empty loading screen. If the host opens the graph before sending data, Mem now tells you the next action to run.
Graph Intelligence is more reliable when you ask about the selected item. Mem now keeps the selected graph object in the chat context even when the canvas sends it through a wrapped event shape.
Conversation distillation now explains the real fix when a language model is missing. Downloading the Search Index model is not enough for distillation; Mem points you to the Default or Agents language model setting instead of showing a raw 422 error.
The update action is clearer after a download finishes. The title-bar update button now asks for the same explicit install confirmation as Settings, and Mem still never installs or restarts from a background download without your action.

Improved

Vision settings now say what actually happens in every language: if you do not choose a remote vision provider, Library uses the on-device local vision model rather than inheriting your default text model.
Long-running Codex and AI Now search cards stay compact while they are still working, then open when results are ready, so a single tool call no longer takes over the whole conversation.
Busy graph canvases get a quieter ontology legend that stays compact until you expand it or interact with a type.

Ready to try Mem?

Download the app, connect your AI tools, and keep useful context where you control it.

Download Read Docs

Changelog

The meaningful changes in Nowledge Mem, written for people using it every day.

Latest release

0.10.67

21. Aug. 2026

0.10 is a ground-up engine rewrite

0.10.67

21. Aug. 2026

0.10.67Latest release21. Aug. 2026

Safer Answers From Your Own Knowledge

Highlights

Retrieved knowledge is now handled as evidence, not instructions.
Feed and Knowledge Answer are more resistant to old content steering a new task.

Fixed

Feed and Knowledge Answer agents now keep retrieved content behind a stronger prompt boundary. Memories, Threads, Library artifacts, and Team workspace material are treated as evidence to read, not commands to follow.
This reduces indirect prompt-injection risk while keeping citations, attribution, and Space visibility boundaries intact.

0.10.66

20. Aug. 2026

0.10.6620. Aug. 2026

AI Now Reviews Its Own Skill Edits, and the App Recovers Instead of Going Blank

Highlights

AI Now can propose Skill updates for your review, and reattaches instead of dead-ending after a dropped connection.
A failed view now shows a retry instead of taking down the whole window.
Search rebuilds keep getting steadier and more precise on large libraries.

Added

AI Now can propose an update to one of your Skills during a conversation. You see a diff before anything changes, and can approve, edit, or decline it inline.
Settings now shows whether the local CLI is correctly wired to your active remote or Access Anywhere connection, and can repair the wiring in one click when it drifts.

Improved

Editing an AI Identity now happens directly in its reading layout — the name, essence, charter, and voice stay at their normal size and position while you edit, instead of jumping into a form.
Search index rebuilds keep getting more memory-bounded and precise on large libraries: rebuild batches stay within a fixed size, importing new conversations no longer forces a full semantic re-embed, and rare source metadata cases no longer stall a rebuild.
Local context rule files now show the same source, scope, and as-of freshness markers as Nowledge Cloud, closing the last gap in the provenance work started in 0.10.65.

Fixed

AI Now no longer dead-ends into "Use Continue" after a dropped connection while a turn is still running on the server. It reattaches to the live turn instead of asking you to restart it.
A failed request or a rendering error in one view no longer blanks the entire window. Memories, Threads, and every other view now confine the failure to that pane and offer a retry.
Remote MCP OAuth pairing now keeps the authenticated public HTTPS origin ahead of any simultaneously running Quick Link, so a Tailscale Funnel pairing can no longer regress to a rotating trycloudflare.com address.
The desktop app no longer spawns netstat.exe while shutting down on Windows, removing a rare source of shutdown hangs and OS error dialogs.
A provider timeout you configure in Settings is now honored end to end for Background Intelligence and AI Now, instead of being silently overridden by a shorter built-in default.

0.10.65

19. Aug. 2026

0.10.6519. Aug. 2026

A Clearer Home for AI Identity and More Reliable Capture

AI Identities and Team knowledge are easier to understand at a glance, while AI Now sessions, coding-agent conversations, and long Grok chats return to Mem more reliably.

Highlights

AI Now can act as the identity you maintain in Mem, then return its work as a durable Thread.
Session capture survives restarts and conflict recovery without duplicating or dropping turns.
Search rebuilds, backup restores, and repeated PDF imports close several long-running reliability gaps.

Added

In the desktop app, AI Now can run a task as a chosen AI Identity. Completed sessions return to Threads automatically, where they stay searchable and can contribute to future Memories.
Devin and LangGraph now have first-class Mem integrations with scoped Memory access and identity-aware conversation capture.

Improved

Automatic session capture resumes from a durable checkpoint after restarts and sends only new turns. If the destination was deleted or changed, Mem performs one bounded reconciliation instead of duplicating or losing the conversation.
Library Wiki topics keep their names and explanations tied to current Memory evidence, making broad or drifting topics easier to understand and refine.
Team members can review Working Memory rulings and the evidence behind contested claims directly from the Feed, without needing workspace-admin access.
In Nowledge Cloud, scheduled briefings now follow the workspace time zone, including daylight-saving changes.
Nowledge Cloud can schedule deferable Background Intelligence for your preferred local hours, helping you use capacity-sensitive or lower-cost model windows without delaying interactive, manual, or safety work.
AI Identities now have one clear home with a stronger presence, readable profiles, and inline editing. Team brings people, shared knowledge, and usage into three focused views, so the same controls no longer compete across the app.
Nowledge Cloud keeps Threads, Memories, and search results inside the workspace you are viewing. When a personal workspace has no content yet, Mem explains that Team knowledge lives in a separate workspace instead of showing an ambiguous empty screen.

Fixed

Large search-index rebuilds no longer remain trapped in a repeated Message FTS repair loop, especially in long-running Docker libraries.
Backup imports restore Thread and Source content before rebuilding graph links, so restored records no longer point at missing payloads.
Re-uploading the same scanned PDF now recovers its existing Library source instead of failing as a duplicate.
MCP App search cards no longer show a stale graph after the selected result changes, and AI Now artifact previews open reliably.
The browser extension now captures complete Grok conversations opened inside X, including turns outside the visible page, and preserves comparison tables with their rows, columns, links, and code intact. Sync, Distill, and Import restore your scroll position and refuse to save when completeness cannot be verified.
Provider settings now describe model routing as it actually works: most Background Intelligence uses Default, while AI Now, Timeline chat, and other interactive agents use Agents unless you choose Same as default.
The desktop update confirmation now follows your selected app language, including its version details and install or cancel actions.
Connection screens now show the actual workspace address that each client will use, so local, Access Anywhere, and Cloud targets are no longer easy to confuse.
Credit top-up links now open reliably from the desktop app instead of being blocked by the app window.
The desktop sidebar now keeps account sign-out within reach even when the web account menu is unavailable.
The desktop app now verifies that the nmem command actually runs from a new terminal before marking the CLI as installed. When PATH setup is needed, it shows the exact change and asks before applying it.

community-2026-08-18

18. Aug. 2026

community-2026-08-1818. Aug. 2026

Nowledge Mem for Devin

Devin teams can now carry durable context across CLI, Desktop, and enterprise Cloud sessions without mixing account identity with agent memory.

Platform

community-2026-08-17

17. Aug. 2026

community-2026-08-1717. Aug. 2026

Nowledge Mem for LangGraph

A first-class Python connector gives LangGraph agents shared memory without taking over their execution state.

Highlights

One durable memory layer for LangGraph agents, with identity and Thread semantics that survive model and deployment changes.

Platform

LangGraph agents can now load the right Nowledge Agent and Space context on every turn, use scoped memory tools, and keep completed conversations as searchable Mem Threads.
The connector keeps LangGraph checkpoints untouched, treats assistant IDs as deployment metadata rather than conversation identity, and captures a parent/subagent run once instead of duplicating every subgraph.

0.10.64

16. Aug. 2026

0.10.6416. Aug. 2026

Safer PDF Imports and Smoother Setup

Highlights

Searchable PDFs avoid redundant OCR and stay usable when optional vision work fails.
Unsaved Memory edits survive background refreshes and failed saves.
Remote setup, Feed links, and web copy actions are steadier across everyday browser setups.

Fixed

Unsaved Memory edits now survive detail refreshes, failed saves, and changes made while a save is still in flight.
Searchable scanned PDFs now keep their existing text instead of sending every page through OCR. If optional figure analysis fails, the document remains searchable and reports a partial result rather than a failed import.
Remote MCP OAuth setup now keeps the stable public HTTPS hostname that opened the pairing flow, and ChatGPT Subscription no longer offers a Codex-only model that the local runtime cannot execute.
Working Memory links from the Feed now open the exact Memory they point to, including memories that belong to another Space.
Copy buttons in the web and Cloud app now keep working in restricted browsers, local network pages, and other places where the standard clipboard API is unavailable.
Firefox extension setup now keeps the permission prompt attached to your click, so connecting the extension to Mem is more reliable.
Provider setup errors now keep the useful provider detail instead of replacing it with a generic model-capability hint.
The Copilot CLI community plugin now works cleanly on Windows PowerShell.

0.10.63

15. Aug. 2026

0.10.6315. Aug. 2026

Faster Starts and More Reliable Imports

Highlights

Web cold starts download about one-sixth of the previous entry payload.
Long scanned PDFs preserve useful partial OCR results.
Kimi Code imports stay stable across transcript rewrites.

Improved

The web app now paints its loading shell before the API bootstrap completes and downloads only the active language on first load. The initial entry payload drops from about 1.4 MB to 258 KB gzip, so cold starts feel immediate instead of blank.
MiniMax can now be selected as the Vision provider for Library PDF and image parsing. Mem routes vision requests to MiniMax's compatible endpoint even when chat uses its Anthropic-style endpoint.

Fixed

ChatGPT custom MCP connections no longer finish OAuth successfully and then fail with HTTP 422 and zero available actions when the host starts without an MCP session ID.
Long scanned PDFs now keep the pages that fit within the OCR budget instead of failing the entire Library import. Source metadata clearly records partial, skipped, failed, and budget-exhausted work.
Kimi Code session imports now keep stable message identities when transcript files are rewritten, and discovered sessions preserve their Kimi source and project or workspace identity instead of being grouped incorrectly.
Running out of configured graph memory no longer traps Mem in a fatal startup loop. The service stays available in a degraded state so you can raise the limit or export diagnostics, and automatic sizing records a safer floor for the next start.

0.10.62

13. Aug. 2026

0.10.6213. Aug. 2026

DeepSeek Harness, ZCode, and Safer Backups

Highlights

DeepSeek Harness and ZCode are now first-class Mem integrations.
Backups restore Library original files, not just parsed text.
Long remote-embedding rebuilds are bounded before they reach your provider.

Added

Fixed

Cloud restore now brings Library originals back with the backup. PDFs, documents, images, and other uploaded files can be previewed or downloaded after restore instead of leaving only parsed text behind.
Search Index rebuilds no longer send oversized Memory or Library text straight to a remote embedding provider. Long items are bounded before the request, so one large note is much less likely to keep semantic search stuck in text-only mode.
ChatGPT custom MCP connector setup is more reliable when OAuth discovery is probed after Allow. The pairing page now answers the compatibility endpoint without widening the connector's default read-only scope.
Data Transfer and Library recovery now fail more explicitly: recoverable export and restore mismatches become warnings, while binary-looking Library uploads are quarantined instead of being indexed as broken placeholder text.
AI Now no longer automatically replays slow or interrupted model requests. If a local or self-hosted model is still working, Mem now stops and lets you continue deliberately instead of queuing the same large prompt again.
The Threads source filter now treats case variants as the same source. Sessions from Hermes, hermes, and related Hermes agent names appear under one clean source instead of duplicate filters.
Desktop installers and self-hosted Docker images are now built with a more resilient PDF parser dependency download, so brief network interruptions during release packaging are less likely to delay an update.

cloud-2026-08-13

13. Aug. 2026

cloud-2026-08-1313. Aug. 2026

A Better Home for Teams on Nowledge Cloud

A wave of improvements for team workspaces on Nowledge Cloud. They arrive with your workspace, so there is nothing to install or update.

Highlights

Every Mem you belong to is one switch away, with Personal and Team clearly told apart.
Team imports explain who can read what before you commit.

Added

See every Mem you belong to and switch between them. Workspace pickers now tell Personal from Team at a glance, so you always know whose memory you are looking at.
Name your workspace. Rename it right from the switcher, and a new team is invited to pick its name the first time it opens.
Plan and seats in one place. Your team's plan, seat usage, and a direct path to billing now live together on the Team page.

Improved

Signing in takes you to your team first, so shared work is one step closer.
Application and sign-in emails now arrive in your language and address your company by name.
Invites tell the truth. If an invite email could not be delivered, Mem says so clearly, instead of leaving a teammate waiting on a message that never comes.
Importing into a team workspace now explains up front who will be able to read what, before anything is brought in.

0.10.59

13. Aug. 2026

0.10.5913. Aug. 2026

Clearer Setup and Graph Feedback

Highlights

Graph Explorer no longer gets stuck on an empty load.
Distillation tells you when a language model is missing.
Update install stays explicitly user-confirmed.

Fixed

Graph Explorer inside MCP-capable clients no longer sits on an empty loading screen. If the host opens the graph before sending data, Mem now tells you the next action to run.
Graph Intelligence is more reliable when you ask about the selected item. Mem now keeps the selected graph object in the chat context even when the canvas sends it through a wrapped event shape.
Conversation distillation now explains the real fix when a language model is missing. Downloading the Search Index model is not enough for distillation; Mem points you to the Default or Agents language model setting instead of showing a raw 422 error.
The update action is clearer after a download finishes. The title-bar update button now asks for the same explicit install confirmation as Settings, and Mem still never installs or restarts from a background download without your action.

Improved

Vision settings now say what actually happens in every language: if you do not choose a remote vision provider, Library uses the on-device local vision model rather than inheriting your default text model.
Long-running Codex and AI Now search cards stay compact while they are still working, then open when results are ready, so a single tool call no longer takes over the whole conversation.
Busy graph canvases get a quieter ontology legend that stays compact until you expand it or interact with a type.

Ready to try Mem?

Download the app, connect your AI tools, and keep useful context where you control it.

Download Read Docs