AI memory comparison

AI memory compared: Tephra and ten other memory systems

These eleven products all try to stop an AI from forgetting you, and they are built for different people. Mem0, Supermemory, Zep, Letta and Cognee are infrastructure for people building agents, though several of them also ship an MCP server an ordinary user can connect. Basic Memory and Pieces are finished products for individuals: one keeps your memory as Markdown files you own, the other captures your work automatically and keeps it on your machine. ChatGPT memory, Claude memory and Gemini memory are features inside one vendor's app. Tephra is a hosted memory that the AI apps you connect share. The table shows each one's own description of itself with a source for every competitor cell, and the sections below say plainly where each competitor is the better choice - including the several places where they beat us.

Sources read and prices checked 2026-10-10.

Obsidian Metrics makes Tephra and Tracker, the two products this page compares against others. This page is written by the maker of one of the products compared, so read the competitor columns against their own pages, linked below each table. Competitor facts come only from the vendor pages listed under Sources, read on the date shown above; vendors change plans and prices without notice. We are not affiliated with any company named here, and all product names belong to their owners.

Side by side

What it is

  • Tephra: A hosted memory your connected AI apps share: they recall from it and write to it over MCP or REST.
  • Mem0: Persistent memory for AI applications, sold as a managed platform and as an open-source library or Docker stack. Aimed at developers building AI apps. [1]
  • Supermemory: Context infrastructure for AI agents: memory, retrieval (SuperRAG) and user profiles through one API. It also offers a remote MCP server that gives MCP-compatible assistants a shared memory layer, with team spaces. [2] [3]
  • Zep: A context layer for enterprise data that organizes business data, documents and conversations in temporal context graphs. Graphiti is its open-source graph framework. [4]
  • ChatGPT memory: A feature inside ChatGPT that can draw on past chats, saved memories, custom instructions, files in Library and, depending on plan and region, content from connected apps such as Gmail. [5]
  • Claude memory: A feature inside Claude that saves memory as a set of individual topics as you chat, with a separate memory space and dedicated summary for each project. [6]
  • Letta: A platform for stateful agents: in its own words, "AI with advanced memory that can learn and self-improve over time". The open-source harness gives an agent a git-versioned memory filesystem, skills, schedules and background subagents that reorganise memory. [7] [8]
  • Cognee: In its own words, "the open-source AI memory platform for agents". Its MCP server gives assistants such as Claude Code, Cursor and Codex persistent memory backed by a knowledge graph. [9] [10]
  • Basic Memory: A knowledge base you own: a knowledge graph written in plain Markdown files, reachable by AI assistants over MCP. Its pitch against built-in memory is that yours is "large, structured, and inspectable" and the files stay yours. [11] [12]
  • Pieces: "The memory layer for modern work." A desktop app plus a local service whose Long-Term Memory engine continuously captures context from your workflow - code you copy, screens you view, audio you hear - and stores it on your own device. [13] [14]
  • Gemini memory: A personalization feature inside Google's Gemini apps. It can draw on the memory of your past Gemini chats and on Connected Apps, with saved facts branded Personal Intelligence. [15] [16]

Entry price

  • Tephra: Tephra needs a paid plan: Premium at $19 a month in the Obsidian Metrics Skool community or $29 a month direct, or a $399 one-time seat. No free tier.
  • Mem0: Hobby is free (10K add requests and 1K retrieval requests a month). Starter is $19 a month, Pro is $249 a month, Enterprise is custom. [17]
  • Supermemory: Free at $0 with $5 of monthly credits. Pro is $19 a month, Max is $100 a month, Scale is $399 a month, Enterprise is custom. [18]
  • Zep: Free plan with 10,000 credits a month. Flex is $125 a month, Flex Plus is $375 a month, Enterprise is custom. [19]
  • ChatGPT memory: The help page says memory features and controls vary by plan, region, platform and workspace settings. It gives no memory price. [5]
  • Claude memory: Memory is on by default for Free, Pro and Max plans. Team and Enterprise owners decide whether it is available, and it is off for each member until they turn it on. Anthropic states memory is not available to organisations with HIPAA, public-sector or custom data-retention agreements. [6]
  • Letta: A free $0 plan limited to 3 stateful agents, bring-your-own API keys. Pro is $20 a month with up to 20 stateful agents. An API plan is $20 a month with unlimited agents plus $0.10 per active agent per month and $0.00015 per second of tool execution. Teams Pro is $20 per seat. Read 2026-10-10. [20]
  • Cognee: A free plan it calls "free forever": 1 workspace and 1M tokens, no card required, with unlimited users and unlimited API calls. Standard is $1.00 per 1M tokens processed plus $5 per additional workspace. Enterprise is quoted. Read 2026-10-10. [21]
  • Basic Memory: Free and open source to run on your own machine. Cloud Team is $15 per seat a month ($13 at 10+ seats, $12 at 20+), Business is $30 per seat a month. Every cloud plan starts with a 7-day trial. Read 2026-10-10. [22]
  • Pieces: No free tier. Its own pricing FAQ asks "Is Pieces free to use?" and answers "No." Pro is $18.99 per user a month and Enterprise $22.99 per user a month, both billed monthly, with a 7-day trial and a 14-day money-back guarantee. Read 2026-10-10. [13]
  • Gemini memory: Memory of past Gemini chats needs only a personal Google Account with Keep Activity on, at no stated extra cost. Managing saved facts requires a paid Google AI plan. We did not record a Google AI plan price because the figures are rendered in the browser and we could not read them from the page itself. [15] [23]

How usage is limited

  • Tephra: No usage metering. There are only abuse limits: 60 requests a minute per key and 120 a minute per member.
  • Mem0: Metered by add and retrieval requests per month on each plan; usage-based pricing is also offered. [17]
  • Supermemory: Monthly credit balance on every plan, drawn down at a token-based rate card for memory, retrieval and operations. [18]
  • Zep: Monthly credits spent on the data you send (each Episode); overage is billed per credit block. The page also lists per-minute rate limits and project limits by plan. [19]
  • ChatGPT memory: The help page says there is no separate storage limit for what ChatGPT can reference through chat history. [5]
  • Claude memory: Not stated on the page we read. [6]
  • Letta: Mixed: a cap on the number of stateful agents on personal plans (3 free, 20 on Pro), quota plus pay-as-you-go overage on Letta Auto, per-seat on Teams, and metered credits on developer plans, where credits cover LLM inference and CPU. Tool execution is metered per second. [20]
  • Cognee: Metered on tokens processed rather than requests or seats. Seats are explicitly not metered on the free plan and API calls are explicitly unmetered; extra workspaces are a fixed $5 a month each. [21]
  • Basic Memory: Per active seat on cloud, with volume breaks; notes and version history are "unlimited" subject to fair use. Only its agent-infrastructure tier is credit-metered. Running it locally is unmetered. [22]
  • Pieces: Per seat and otherwise unmetered: Pro lists "Unlimited MCP" and lets you switch between models under one plan. No token or request metering is documented. [13]
  • Gemini memory: Consumer subscription tiers rather than metered memory, but with hard eligibility gates: 18 or over, a personal Google Account only, and saved facts are US-only. Google states the feature is not available on work, school or supervised accounts. [15] [23]

Reach across AI apps

  • Tephra: One memory for every AI you connect, led by OAuth sign-in: the Connect page has steps for Claude (web, Desktop, mobile), ChatGPT, VS Code Copilot, Cursor, Claude Code, Gemini CLI, Windsurf, Zed, Mistral Le Chat and Perplexity, plus generic steps for any other MCP app, and a personal key as a fallback, plus REST. AI tools that support neither MCP nor REST calls cannot connect. Tephra's OAuth discovery endpoints were verified live on 2026-10-10.
  • Mem0: Documentation lists setup guides for 22 tools, including LangChain, CrewAI, LlamaIndex and the Vercel AI SDK, and coding-agent guides starting with Claude Code, then Cursor, Codex and others. [1]
  • Supermemory: Remote MCP server that any compatible client can add; the client opens Supermemory in a browser for OAuth sign-in, so no API key is required. Tools include search, profile, save and forget. [3]
  • Zep: The free plan lists 1 Memory MCP Server seat and paid plans list more (5 on Flex, 15 on Flex Plus). Separately, the documentation offers a Claude Code and Codex plugin that bundles a skill with the Zep docs MCP server - that one is documentation, not memory. [19] [4]
  • ChatGPT memory: The controls described are inside ChatGPT. The help page does not describe sharing that memory with other AI apps. [5]
  • Claude memory: Memory is scoped to Claude, with a separate memory space per project, and is shared between Claude chat and Claude Cowork when Cowork runs in the cloud - local Cowork sessions do not use memory. Claude also supports moving memory by pasted text: import from other AI providers (Settings, Memory, Start import) and export of your memory for backup or migration. The help pages call memory imports experimental and still in active development, and list imports for Free, Pro, Max and Team plans; the export instructions carry no such label, and core memory is not described that way. [6] [24]
  • Letta: A hosted remote MCP server at api.letta.com/mcp that any MCP client can reach, and it names Claude Code and Cursor. Authentication is a bearer token only: its docs state plainly that "OAuth-only clients are not supported yet". [25]
  • Cognee: An MCP server you run yourself, in Standalone Mode with its own local database or in API Mode against a shared Cognee backend, naming Claude Code, Claude Desktop, Cursor, Cline, Continue and Codex. We found no hosted remote MCP URL and no OAuth documented, so a user must run a process and supply a token; its docs state that Cognee MCP and Cognee Cloud are separate systems with different APIs and authentication schemes. [10]
  • Basic Memory: The widest reach of any competitor here. Hosted MCP at cloud.basicmemory.com/mcp with OAuth and no local install, published in the Claude Directory so it installs across Claude web, desktop, mobile, Claude Cowork and Claude Code, plus a ChatGPT app. API keys are available as an alternative. [12]
  • Pieces: Machine-bound: the MCP server is PiecesOS running on your own computer, which its docs call a required dependency. It names Claude Desktop, Claude Code, Claude Cowork, Cursor, Codex CLI, GitHub Copilot, VS Code, Zed, Windsurf, JetBrains, Raycast and more. There is no hosted endpoint and no OAuth; its docs do describe an ngrok setup to "share PiecesOS for remote access", so remote use means tunnelling your own machine rather than a vendor-hosted server. [26] [14]
  • Gemini memory: None. Google scopes it to the Gemini mobile and web apps, Gemini in Chrome and Gemini on a smartwatch. There is no MCP server and no third-party client; its reach runs inward, through Connected Apps, instead. [15]

Seeing and correcting what is stored

  • Tephra: A member can view memories, edit or delete them, read edit history, contest a memory, export their own data (app session), and revoke any AI key with an activity log.
  • Mem0: The docs navigation links to a developer dashboard; the landing page we read carries little else. Controls for end users are not described on the pages we read. [1]
  • Supermemory: The MCP tools can list memories and documents and save or forget information. Controls for end users beyond that are not described on the pages we read. [3]
  • Zep: The overview describes governance features (access control, data-source tracing, activity review) and the Enterprise plan lists audit logs. Per-user memory editing is not described on the pages we read. [19] [4]
  • ChatGPT memory: Settings, Personalization, Memory: a memory summary you can correct, saved memories you can edit or delete, and sources shown under responses. Temporary chats do not create or update memories. OpenAI may retain logs of deleted saved memories for up to 30 days. [5]
  • Claude memory: Settings, Memory lists topics you can read, edit or delete, and you can tell Claude to remember, change or forget. Memory can be paused or reset, and turned off for a single chat from the plus menu before the first message. Incognito chats are not saved to memory. [6]
  • Letta: Strong. Memory is a git-backed filesystem the agent and you can both inspect and edit, with a memory viewer in the desktop app, a /doctor command that audits placement, duplication and token usage, and a backup taken before memory is restructured. [8]
  • Cognee: A web UI in its hosted cloud for exploring the knowledge graph, running searches and managing datasets, and a first-class forget operation beside remember and recall. Conflict resolution and per-answer provenance are sold as enterprise features. We did not confirm per-memory version history. [27] [21]
  • Basic Memory: The strongest in this table. Memory is Markdown you can open in any editor or in Obsidian; cloud adds a note editor, per-note file history with restore, point-in-time snapshots with rollback, comments and suggested edits, and documented deletion semantics for both local and cloud copies. [12] [22]
  • Pieces: Good on control, less clear on editing. There is a browsable timeline of captured activity, you can pause capture for a set period, disable long-term memory entirely, exclude individual apps as sources, and clear all stored memory in one action. Its docs note that excluding a source stops new capture but retains what was already captured. We did not confirm per-memory editing or version history. [14] [13]
  • Gemini memory: Still the weakest here, by Google's own description. There is an on/off switch and an editable instructions list, and you can ask Gemini to remember a fact or correct a wrong one conversationally - Google says "Gemini will update its memory based on your correction". But there is no list of remembered facts to read: to delete something Gemini remembered, Google says to delete all chats containing it, and to disconnect the app as well if it came from a connected app. Google also warns that deletion can lag, that connected-app deletions "may not be affected until days later", and that instructions to forget or avoid a topic "doesn't always work perfectly". [23] [15]

Hosting and self-hosting

  • Tephra: Hosted service run by Obsidian Metrics. Self-hosting is not part of the product. We make no encryption-at-rest or compliance-certification claim.
  • Mem0: Open-source version runs as a library or Docker stack on your infrastructure, or use the managed platform. Enterprise lists on-prem deployment, audit logs and SSO; the site shows SOC 2 Type I and HIPAA Ready badges. [1] [17]
  • Supermemory: Cloud platform. Self-hosted deployments are available on Scale and Enterprise, and the docs describe a self-hosted single binary. SOC 2 and HIPAA BAA are listed on Scale. [18] [2]
  • Zep: Cloud service, with Enterprise options for cloud with your own encryption keys or deployment in your VPC. Graphiti is open source. Enterprise lists SOC 2 Type II and HIPAA BAA. [19] [4]
  • ChatGPT memory: A vendor-hosted feature of ChatGPT. [5]
  • Claude memory: A vendor-hosted feature of Claude. [6]
  • Letta: Apache-2.0 on letta-ai/letta and self-hosting is first-class, though the current harness source has moved to letta-ai/letta-code, whose licence we have not verified: its docs describe running Letta fully locally or on your own infrastructure with no Letta account required, with all agent state staying on the device, plus a self-hosted always-on app server. [7] [28]
  • Cognee: Apache-2.0, and the UI is self-hostable free with no account. Two caveats it discloses itself: its Postgres graph store is released as a demo feature with the production version licensed separately, and its enterprise tier includes a proprietary runtime, so the fastest path is not the open part. Hosted workspaces sleep when idle. [9] [27] [21]
  • Basic Memory: AGPL-3.0 and local-first by default, with cloud opt-in per project and a documented mix of the two. Note that AGPL can itself be a blocker for some commercial reuse, though the hosted cloud sidesteps that for an ordinary user. [29] [22]
  • Pieces: Closed source but local-first, which is an unusual pairing. Its docs state that the on-device engine "runs entirely on your device" and that "All captured data is processed and stored locally. Nothing leaves your machine unless you explicitly choose to share it." We found no compliance certification on the pages we read. There is no self-hosted server to run and no licence to inspect. [14] [13]
  • Gemini memory: Cloud only and closed, as a consumer feature of Google's own apps. No self-hosting and no data-residency option described. [15]

How memories get in

  • Tephra: The connected AI writes them by calling a remember tool. Tephra does not capture chats automatically and has no Gmail, Slack or Drive connectors. You can upload text, markdown, JSON, XML and text-layer PDF files.
  • Mem0: Developers save and search memories through the API or library. [1]
  • Supermemory: Connectors for Google Drive, Gmail, Notion, OneDrive, GitHub, Granola and a web crawler (the pricing page lists Gmail on Max and S3 and web crawler connectors on Scale). Ingests text, chats, PDFs, images, video and code. [30] [18] [2]
  • Zep: Consolidates business data, documents and conversations into the context graph. [4]
  • ChatGPT memory: From past chats, saved memories (details you ask it to remember or that it saves), custom instructions, Library files and, depending on plan and region, connected apps. [5]
  • Claude memory: Claude saves on its own and you can tell it to remember something. Government ID numbers, criminal history, financial account numbers and immigration status are never saved. [6]
  • Letta: Automatically, explicitly, and in the background: the agent updates memory when it learns something durable, you can teach it directly, and background subagents review recent conversations to consolidate lessons without interrupting you. It also connects to messaging channels; the repository we read names Slack, Telegram, Discord and custom channels. [8] [7]
  • Cognee: An explicit remember call or an ingestion pipeline, plus real connectors: its cloud reads Slack, GitHub and Google Drive into a shared workspace memory, and Slack, Notion, Linear and Google Drive integrations are listed on its paid tier. [27] [21]
  • Basic Memory: Explicit tool calls from the assistant, writing the Markdown yourself in any editor, and importing project archives or Claude and ChatGPT exports. Automatic capture appears only as a "Harness Capture" entry in the cloud docs navigation, with no description in the body text we read. We found no Gmail, Slack or Drive connectors. [12]
  • Pieces: Fully automatic and continuous, with no tool call at all: its engine captures code you copy, screens you view and audio you hear, and its sources include browser history, email, notes and meetings. It also has a Google Calendar connector. [14]
  • Gemini memory: Both, and broadly. Gemini can learn from your chats automatically when Keep Activity is on, you can tell it to remember something explicitly, and Connected Apps let it draw on your Google apps. [15] [23]

Where the others are better

  • Mem0. You can run the open-source version on your own infrastructure, and there is a real free plan for prototyping. Tephra has neither. [1] [17]
  • Supermemory. It has data connectors (Google Drive, Gmail, Notion, OneDrive, GitHub and more; Gmail is listed on Max), an OAuth MCP server with team spaces, and a free tier. Tephra has no member connectors and no free tier. [30] [3] [18]
  • Zep. It targets enterprise requirements: temporal context graphs, governance, and SOC 2 Type II and HIPAA BAA on the Enterprise plan. Tephra makes no such claims. Zep organizes data in temporal context graphs, and Graphiti builds temporal knowledge graphs, while Tephra's stale-fact handling is partial. [4] [19]
  • ChatGPT memory. It needs no setup if you already use ChatGPT, the controls live in the app you are already in, and it can draw on files and, depending on plan and region, connected apps such as Gmail. Tephra has no member connectors. [5]
  • Claude memory. It is built into Claude with a separate memory space per project, a per-chat switch and incognito chats, needs no setup, and can move memory by pasted text, with an import path its help pages call experimental. [6] [24]
  • Letta. Is Apache-2.0 on letta-ai/letta - the current harness has moved to letta-ai/letta-code, whose licence we have not verified - and runs fully locally with no account, so it clears a compliance or air-gap review that Tephra cannot. It has a real $0 tier, it writes durable memory without waiting for a model to call a save tool, and background subagents re-read past conversations to consolidate and de-duplicate memory - where Tephra leaves correcting stale facts to you. Its memory is a git repository you can diff and revert in your own editor, and it reaches messaging channels including Slack, Telegram and Discord, which Tephra does not. [7] [20] [8]
  • Cognee. Lets a buyer start with no card and no sales call on a free plan that is not a token: unlimited users and unlimited API calls. It is Apache-2.0 including a self-hostable UI, it prices at a flat $1.00 per 1M tokens that anyone can model on a spreadsheet, and it ingests Slack, Google Drive, GitHub, Notion and Linear - so its memory fills from where work already happens, while Tephra's fills only when an AI calls its save tool. It also sells conflict resolution and per-answer provenance, which Tephra asks you to do by hand. [21] [27] [9]
  • Basic Memory. Matches Tephra's hosted-MCP-plus-OAuth story and then adds distribution Tephra has not got: a Claude Directory listing that installs across Claude web, desktop, mobile, Claude Cowork and Claude Code, plus a ChatGPT app. Its transparency beats ours on the merits rather than the marketing, because memory is Markdown on disk with per-note version history, snapshot rollback and suggested edits. It is free forever locally under AGPL-3.0, so there is no cost to try and a clear answer to "what if the vendor disappears", and exit is a file copy rather than a migration. Its cheapest paid tier, $15 a seat, undercuts ours, and it offers SSO/SAML and HIPAA-compliant hosting at $30. [12] [22] [29] [11]
  • Pieces. Captures without asking anything of the model or the user. Tephra holds only what an AI decided to save; Pieces holds the screens you looked at, the code you copied and the meeting you sat in, so for "why did we decide this three weeks ago" it has an answer and Tephra may have nothing. It keeps data on your own machine by architecture rather than policy - its docs say nothing leaves your machine unless you choose to share it - while Tephra's encryption at rest is not confirmed. Its consent controls - a pause for a duration you choose with a "Paused until" label, auto-stop on screen lock, per-app access control, and a clear-data dialog you can filter by time period, modality and app source - are more concrete than an inspect-only view, and it ingests screen, audio, email and meetings, which Tephra cannot. [14] [13]
  • Gemini memory. Is already on, already paid for, and already attached to the user's real life through Connected Apps, so Gmail, Photos and Calendar context arrives with no integration work at all - against which Tephra knows only what an AI typed into its save tool. Memory of past chats costs nothing on a personal Google Account, capture is automatic, correction happens conversationally inside the chat, and it needs no setup on phone, web, Chrome or a watch. [15] [23]

Where Tephra is better

  • Connected AI apps read and write the same memory directly, rather than moving it between apps by pasting text. Claude's import and export, for example, works by pasted text, and its import path is marked experimental.
  • A flat price with no usage metering, instead of credits or request quotas.
  • A shared library that Tephra grows every hour from public research and regulatory sources, included with the product. We did not find a competitor that bundles one; that is a qualitative check, not an exhaustive market survey.
  • Recall in the app tells you which related memories it did not load and how complete the answer looks.

Where Tephra falls short

  • Tephra has no free tier and no self-hosted option.
  • The MCP tools return ranked records only. The completeness alert is shown in the app, not through MCP.
  • No automatic chat capture, and no connectors to Gmail, Slack or Drive.
  • No image or audio ingest, and no OCR for scanned PDFs.
  • Stale-fact handling is partial.
  • No published, vendor-comparable benchmark, and no accuracy lead we can defend.

What we have and have not measured

On 50 held-out LongMemEval-S questions never used in Tephra development, Tephra answered 39 of 50 correctly (78%, 95% confidence interval 65-87%) with a gpt-4o-mini reader, and every evidence session was in its top 10 results for 46 of 46 answerable questions.

Method: LongMemEval-S cleaned, stratified 100-question subset; gpt-4o-mini reader over top-10 retrieved items; official GPT-4o judge prompt; one run, 2026-10-08; held-out questions are the 50 never used in earlier Tephra tuning; competitors run locally from open source under the same harness.

In that same-harness run, none of Tephra's leads over the other systems tested was statistically significant after correcting for multiple comparisons, and its margin over a plain keyword-search baseline was not significant either. We do not claim Tephra is the most accurate memory. We did not benchmark Supermemory, Zep, ChatGPT memory or Claude memory, and we do not compare ourselves with any vendor's self-reported numbers.

Frequently asked

Is Tephra more accurate than Mem0, Zep or Supermemory?

We cannot say so. In our same-harness test none of Tephra's leads over the other systems was statistically significant after correction, and its margin over a plain keyword-search baseline was not significant either. We did not run Supermemory or Zep. Pick on fit, not on accuracy.

Which is the best AI memory for developers building an app?

Mem0 and Supermemory describe themselves as infrastructure for AI agents and apps, Zep as a context layer for enterprise data, and each has a free or low-cost entry plan listed on its pricing page. Tephra is a hosted memory for people connecting their own AI apps, not an SDK for building memory into your product.

Do ChatGPT memory and Claude memory share anything with each other?

Claude's help center describes importing memory from other AI providers and exporting Claude memory as text, and marks the import path experimental. The ChatGPT page we read does not describe a transfer feature. Tephra is designed so that the AI apps you connect read and write one memory directly.

More comparisons

Tephra

Sources

Competitor facts come only from the pages below, accessed 2026-10-10.

  1. Mem0 documentation overview: https://docs.mem0.ai/introduction (accessed 2026-10-10)
  2. Supermemory documentation overview: https://supermemory.ai/docs/intro (accessed 2026-10-10)
  3. Supermemory documentation, Supermemory MCP: https://supermemory.ai/docs/supermemory-mcp/mcp (accessed 2026-10-10)
  4. Zep documentation overview: https://help.getzep.com/overview (accessed 2026-10-10)
  5. OpenAI Help Center, Memory in ChatGPT: https://help.openai.com/en/articles/8590148-memory-in-chatgpt (accessed 2026-10-10)
  6. Claude Help Center, chat search and memory: https://support.claude.com/en/articles/11817273-use-claude-s-chat-search-and-memory-to-build-on-previous-context (accessed 2026-10-10)
  7. Letta source repository (Apache-2.0): https://github.com/letta-ai/letta (accessed 2026-10-10)
  8. Letta documentation, memory: https://docs.letta.com/configuration/memory (accessed 2026-10-10)
  9. Cognee source repository (Apache-2.0): https://github.com/topoteretes/cognee (accessed 2026-10-10)
  10. Cognee documentation, MCP overview: https://docs.cognee.ai/cognee-mcp/mcp-overview (accessed 2026-10-10)
  11. Basic Memory home page: https://basicmemory.com (accessed 2026-10-10)
  12. Basic Memory documentation, cloud guide: https://docs.basicmemory.com/cloud/cloud-guide (accessed 2026-10-10)
  13. Pieces pricing: https://pieces.app/pricing (accessed 2026-10-10)
  14. Pieces documentation, long-term memory: https://docs.pieces.app/products/core-dependencies/pieces-os/long-term-memory (accessed 2026-10-10)
  15. Google Help, personalization with memory of your past Gemini chats: https://support.google.com/gemini/answer/16598469 (accessed 2026-10-10)
  16. Google Help, get personalization in Gemini Apps: https://support.google.com/gemini/answer/16598623 (accessed 2026-10-10)
  17. Mem0 pricing: https://mem0.ai/pricing (accessed 2026-10-10)
  18. Supermemory pricing: https://supermemory.ai/pricing (accessed 2026-10-10)
  19. Zep pricing: https://www.getzep.com/pricing (accessed 2026-10-10)
  20. Letta pricing: https://docs.letta.com/pricing (accessed 2026-10-10)
  21. Cognee pricing: https://cognee.ai/pricing (accessed 2026-10-10)
  22. Basic Memory pricing: https://basicmemory.com/pricing (accessed 2026-10-10)
  23. Google Help, manage what Gemini remembers: https://support.google.com/gemini/answer/16598625 (accessed 2026-10-10)
  24. Claude Help Center, import and export your memory: https://support.claude.com/en/articles/12123587-import-and-export-your-memory-from-claude (accessed 2026-10-10)
  25. Letta documentation, hosted MCP: https://docs.letta.com/platform/hosted-mcp (accessed 2026-10-10)
  26. Pieces documentation, MCP: https://docs.pieces.app/products/mcp (accessed 2026-10-10)
  27. Cognee documentation, Cognee Cloud overview: https://docs.cognee.ai/cognee-cloud/overview (accessed 2026-10-10)
  28. Letta documentation, self-hosting: https://docs.letta.com/self-hosting (accessed 2026-10-10)
  29. Basic Memory source repository (AGPL-3.0): https://github.com/basicmachines-co/basic-memory (accessed 2026-10-10)
  30. Supermemory documentation, connectors overview: https://supermemory.ai/docs/connectors/overview (accessed 2026-10-10)