Claude Import Memory: Availability, Supported Plans, How it Works, and Current Limits

Claude’s import memory feature is one of the more concrete attempts to reduce switching friction between AI assistants, though the way it works is narrower and more controlled than the phrase itself may suggest.

This is not a live account sync between providers.

It is a manual import flow inside Claude that begins with exported memory from another AI service and ends with Claude extracting what it considers key information into its own memory system.

That distinction defines the feature from the start.

It also explains why the feature can be useful without behaving like a perfect transfer of another assistant’s internal state.

Anthropic makes the feature available broadly across Claude plans, which gives it much wider reach than a premium-only migration tool.

At the same time, Anthropic places several clear boundaries around what gets imported, how quickly it appears, and how reliably it will be retained.

The most important limit is simple.

Imported memory is experimental, and Anthropic explicitly says Claude may not always successfully incorporate it.

The second limit is just as important.

Claude’s memory system is oriented toward work-related usefulness, so imported personal details that are unrelated to work may not be retained in the same way.

Once those two boundaries are kept in view, the feature becomes much easier to understand in practical terms.

··········

How Claude’s import memory feature actually works.

Claude import memory is a manual migration flow that converts pasted exported memory from another AI provider into Claude memory edits rather than creating a direct cross-provider sync.

The feature begins outside Claude.

The user first exports memory from another AI provider.

That exported memory is then brought into Claude through Claude’s built-in import flow on the supported surfaces.

Anthropic documents two entry points for the process.

One route is through Settings > Capabilities > Memory > Start import.

The other route is through the home-screen card labeled Import memory to Claude.

Once inside the import flow, the user pastes the exported memory into the provided text box and clicks Add to memory.

Claude then extracts what it considers key information and stores that information as individual memory edits inside Claude’s memory system.

This is a very specific execution model.

Claude is not mirroring another provider’s memory database.

It is interpreting imported text and converting parts of it into its own memory structure.

That design explains both the usefulness of the feature and its limits.

It gives Claude a way to absorb prior context from another assistant without requiring a provider-level integration.

It also means the transfer is mediated by Claude’s own extraction logic rather than by a lossless system-to-system import.

........

· Claude import memory starts with exported memory from another AI provider.

· The imported content is pasted into Claude’s own import flow rather than synced automatically between accounts.

· Claude converts the imported material into individual memory edits inside its own memory system.

........

Import flow structure

Step	What happens
Export	The user exports memory from another AI provider
Entry point	The user opens Claude’s import flow from settings or the home-screen import card
Transfer method	The user pastes the exported memory into Claude
Processing	Claude extracts key information
Result	Claude stores the extracted information as individual memory edits

··········

Which Claude plans and surfaces support memory import.

Anthropic makes the import-memory feature available across the full current Claude plan range, though the documented import surfaces are narrower than the overall Claude product footprint.

Anthropic’s help documentation states that memory import is available for all Claude users, including free, Pro, Max, Team, and Enterprise plans.

That wide plan coverage is one of the strongest parts of the feature’s rollout posture.

It means memory migration is not restricted to high-paying tiers.

A free user moving from another assistant can still use the import flow.

A Team or Enterprise user can do the same, subject to their organization’s broader memory settings.

The supported surfaces are more specific.

Anthropic explicitly documents memory import on the web and Claude Desktop.

That wording is important.

It confirms those two surfaces directly.

It does not, in the reviewed source set, clearly confirm that the import flow itself is also available on Claude mobile.

The broader Claude memory system extends across more of the product, and Anthropic’s release notes discuss memory availability widely, though the import article itself names only web and Desktop for imports.

That is the cleanest way to frame availability without overstating it.

........

· Memory import is available across free, Pro, Max, Team, and Enterprise plans.

· The documented import surfaces are the web app and Claude Desktop.

· The reviewed official sources do not clearly confirm the same import flow on mobile.

........

Plan and surface availability

Area	Confirmed status
Free	Supported
Pro	Supported
Max	Supported
Team	Supported
Enterprise	Supported
Web	Supported
Claude Desktop	Supported
Mobile import flow	Not clearly confirmed in reviewed sources

··········

How the import flow is structured from export to memory edits.

The import feature is designed as a guided conversion process in which Claude extracts and rewrites usable context into its own memory layer instead of storing the raw external export as-is.

Anthropic’s documentation is unusually direct about the sequence.

The user starts by exporting memory from the previous AI service.

Anthropic even provides a suggested export prompt for the older assistant.

That prompt asks the previous system to list stored memory and learned context across several categories such as response instructions, projects, goals, tools, languages, frameworks, personal details, and preferences or corrections, and to return the information in a single code block.

The user can also customize that export prompt.

Anthropic explicitly notes that the prompt can be edited to exclude sensitive information before anything is brought into Claude.

After the user pastes the exported content into Claude’s import interface, Claude does not simply preserve it as one static imported blob.

Instead, Claude extracts key information and turns it into discrete memory edits.

Those edits can then be reviewed through Manage edits.

This design has a practical consequence.

The feature is partly a migration tool and partly a filtering layer.

The user is not only transporting memory.

The user is handing it to Claude for selection and restructuring.

That is why the post-import output is a set of memory edits rather than a literal foreign-memory archive inside Claude.

........

· Anthropic provides a suggested export prompt for the previous AI provider.

· The imported content can be customized before it enters Claude.

· Claude restructures the imported material into individual memory edits instead of storing one raw imported memory block.

........

From export to memory edits

Stage	Operational detail
Export prompt	Anthropic suggests a prompt that requests stored memory and learned context from the old AI service
User editing	The user can remove sensitive information before import
Paste into Claude	Imported text is pasted into Claude’s import box
Claude processing	Claude extracts key information
Stored result	Information appears as individual memory edits reviewable in Manage edits

··········

What Claude tries to retain from imported memory and what it may leave out.

Claude’s memory system is designed to retain work-related context more reliably than unrelated personal detail, which means imported memory is filtered through Claude’s own usefulness criteria.

Anthropic states that Claude’s memory is designed to focus on work-related topics in order to improve collaboration.

That single line has major practical implications.

It means imported memory is not handled as a neutral archive where every detail has equal status.

Claude is optimized to preserve information that improves future working interactions.

Anthropic also states that Claude may not retain personal details unrelated to work.

So even if another AI provider has been storing broad personal context, Claude may not preserve the same material after import.

This is one of the clearest scope boundaries in the whole feature.

The import flow is not promising comprehensive identity migration.

It is promising selective incorporation into Claude’s memory model.

Anthropic also gives users a fallback route.

If there is specific information the user wants stored and it does not appear after import, Anthropic directs the user to Settings > Capabilities > View and edit your memory to add it manually as an edit.

That confirms the underlying logic.

Claude import memory is useful as a first-pass transfer layer, though manual follow-up remains part of the product design when certain details need to be pinned more explicitly.

........

· Claude prioritizes work-related memory rather than treating all imported content equally.

· Personal details unrelated to work may not be retained after import.

· Anthropic provides manual memory editing as the fallback path when important information does not survive the automated import pass.

........

Retention logic after import

Type of imported information	Likely treatment inside Claude
Work-related preferences and context	More aligned with Claude’s memory design
Project and goal information	Strong fit for memory retention
Tool, language, and framework information	Strong fit for memory retention
Personal details unrelated to work	May not be retained
Missing but important details	Can be added manually through memory editing

··········

Why imported memory is not the same as a direct account sync.

The feature moves context between assistants, though it does not move the original provider’s memory system, ranking logic, or hidden internal structure into Claude.

The phrase import memory can easily suggest something more absolute than what Anthropic is actually offering.

A direct account sync would imply some kind of provider-to-provider linkage in which the source system’s stored memory is transferred in structured form into Claude with very high fidelity.

That is not the documented model.

The documented model is manual export, manual paste, Claude-side extraction, and Claude-side rewriting into memory edits.

This means the imported result depends on at least three filters.

First, whatever the previous provider is willing or able to export.

Second, whatever the user chooses to include or exclude before pasting it into Claude.

Third, whatever Claude decides to extract and retain.

That layered structure is why the feature can help users avoid starting from zero while still falling short of a perfect migration.

Anthropic makes this limit explicit by labeling the feature experimental and by stating that Claude may not always successfully incorporate imported memories.

That sentence removes any basis for treating the feature as a deterministic memory clone.

It is closer to an assisted migration layer than to a true account-state replication system.

........

· Claude import memory is a manual migration flow, not a live provider-to-provider sync.

· The final imported result is filtered by the source export, the user’s edits, and Claude’s own extraction logic.

· Anthropic explicitly says imported memories may not always be fully incorporated.

........

Import memory versus direct sync

Area	Claude import memory
Sync type	Manual import, not live sync
Source dependency	Depends on what the previous AI exports
User control	User can edit imported content before submission
Claude control	Claude extracts and stores selected information as memory edits
Fidelity guarantee	Not guaranteed

··········

How long imported memory takes to appear and how it can be reviewed.

Anthropic gives a concrete post-import timing window and also provides a built-in review layer so the user can inspect what Claude actually kept.

Once the import is complete, Anthropic states that the updated memory should appear within 24 hours.

That timing matters because the effect is not necessarily immediate in every session.

The feature should therefore be understood as an asynchronous memory update inside Claude’s broader personalization system rather than as an instant full-state rewrite visible the second the text is pasted.

Anthropic also states that users can review the imported results through Manage edits.

This is important for two reasons.

First, it gives the user a direct way to see what Claude extracted from the pasted import.

Second, it provides a quality-control layer for a feature that Anthropic itself describes as experimental.

That review mechanism reduces ambiguity.

The user does not need to guess whether Claude absorbed the memory in a usable form.

The imported information should surface as memory edits, which can then be checked and, where necessary, adjusted or supplemented through Claude’s broader memory controls.

........

· Anthropic says imported memory should appear within 24 hours.

· The imported result can be reviewed through Manage edits.

· The review layer is especially important because the feature is explicitly experimental.

··········

How Claude’s broader memory system changes the value of imported memory.

Import memory is part of Claude’s larger memory and chat-history framework rather than an isolated migration trick, which changes how useful the feature becomes after the transfer is finished.

Anthropic’s broader memory documentation shows that imported memory feeds into the same memory system that users can toggle on or off in Settings > Capabilities.

This means imported memory is not trapped inside a one-time migration screen.

It becomes part of Claude’s ongoing personalization layer.

Anthropic also states that when Claude references previous conversations, users see citations back to the original chats.

That broader design gives the import feature more operational value than a static import-only archive would have.

Once the migrated information has been incorporated, it can influence future collaboration inside Claude’s existing memory framework.

That turns the feature into a bridge between prior assistant history and future Claude usage rather than a one-off transfer with no lasting integration.

At the same time, this also means imported memory inherits the governance and configuration logic of Claude’s broader memory system, including user-level toggles and organization-level controls where those apply.

........

· Imported memory becomes part of Claude’s broader memory system rather than remaining isolated.

· Claude’s memory controls and chat-history behavior shape the value of imported content after migration.

· The feature is more useful because it feeds into ongoing personalization rather than stopping at one static transfer event.

··········

Which privacy, governance, and admin controls apply to imported memory.

Imported memory sits inside Claude’s normal memory governance framework, which means user controls and enterprise controls both remain relevant after the transfer.

At the user level, Anthropic states that memory and past-chat search can be toggled on or off through Settings > Capabilities.

That means imported memory does not exist outside the normal Claude memory control structure.

For enterprise environments, Anthropic adds stronger organization-level controls.

Owners and Primary Owners can manage memory features at the organization level through Organization settings > Capabilities.

Anthropic also states that disabling memory at the organization level immediately deletes all existing memory synthesis data for all users in that organization.

That is a significant governance rule.

It means imported memory is not only personal configuration.

It is also subject to centralized administrative policy in managed environments.

Anthropic further states that memory synthesis is encrypted at rest and tied to the underlying conversations, while chat summaries are stored alongside conversation data under existing retention policies.

These details show that imported memory is governed inside the same control architecture as Claude’s broader memory system.

........

· Imported memory follows the same user-level memory controls as the rest of Claude’s memory system.

· Enterprise admins can control memory organization-wide.

· Disabling memory at the organization level deletes existing memory synthesis data for that organization.

........

Governance and admin controls

Area	Confirmed status
User memory control	Memory and past-chat search can be toggled on or off
Enterprise admin control	Owners and Primary Owners can manage memory features org-wide
Effect of org-level disable	Existing memory synthesis data is deleted
Data handling	Memory synthesis is encrypted at rest
Storage linkage	Memory synthesis is tied to underlying conversations

··········

What the current limits of Claude import memory are.

The current feature is broad in plan availability, though narrow in determinism, surface confirmation, and guaranteed retention fidelity.

The first limit is reliability.

Anthropic explicitly states that the feature is experimental and still in active development.

It also states that Claude may not always successfully incorporate imported memories.

The second limit is retention scope.

Claude may not retain imported personal details unrelated to work.

The third limit is surface certainty.

The reviewed official sources clearly confirm the feature on web and Claude Desktop, though they do not clearly confirm the same import flow on mobile.

The fourth limit is technical transparency.

Anthropic does not publish, in the reviewed source set, a formal import schema, a maximum paste size, or a deterministic parsing specification for imported memory text.

So the feature should not be described as a fully standardized migration protocol.

It is better described as a broad user-facing import tool with real usefulness and clear practical constraints.

That is a stronger and more accurate description than either extreme.

It is not a superficial gimmick.

It is also not a perfect memory-cloning system.

........

· The feature is experimental and still in active development.

· Imported memory may not always be fully incorporated.

· Web and Desktop support are confirmed, while mobile import availability is not clearly confirmed in the reviewed sources.

· Anthropic does not publish a full formal technical import spec in the reviewed material.

........

Current limits of Claude import memory

Limit area	Current position
Incorporation reliability	Not guaranteed
Development state	Experimental
Retention completeness	Selective rather than universal
Confirmed import surfaces	Web and Claude Desktop
Mobile import confirmation	Not clearly confirmed
Formal import schema	Not clearly published in reviewed sources

··········

EXECUTION CONTRACT AND FEATURE POSTURE

Claude import memory is a manual migration feature inside Claude’s broader memory system, not a live cross-provider sync layer and not an API-based memory-transfer protocol.

The feature exists to let a user move usable memory context from another AI provider into Claude without rebuilding everything from zero by hand.

That is the practical role.

The transfer itself is narrower than the name may initially suggest.

Anthropic documents a workflow in which the user first exports memory from another AI service, then pastes that exported material into Claude’s import flow, after which Claude extracts what it considers key information and stores it as individual memory edits.

This means the feature is not copying another provider’s memory system into Claude in a structured one-to-one way.

It is taking exported text, interpreting it, and converting selected parts into Claude’s own memory layer.

That execution contract explains both the value of the feature and its most important limits.

It can preserve useful prior context.

It does not promise lossless migration.

It also means the quality of the result depends on three separate layers at once, namely what the previous provider exports, what the user pastes into Claude, and what Claude chooses to extract into memory.

........

· Claude import memory is a manual migration flow rather than a live account sync.

· Imported content is processed by Claude and rewritten into memory edits.

· The feature transfers usable context, not the original provider’s full memory structure.

........

Execution posture of the feature

Area	Confirmed position
Core function	Manual transfer of memory context from another AI provider into Claude
Sync model	No live provider-to-provider sync
Processing model	Claude extracts key information from pasted exported memory
Stored result	Individual memory edits inside Claude
Product layer	Part of Claude’s broader memory framework

··········

PLAN COVERAGE, SUPPORTED SURFACES, AND WHAT IS ACTUALLY CONFIRMED

Anthropic has rolled the feature out broadly at the plan level, though the confirmed import surfaces are narrower than Claude’s full product footprint.

The plan coverage is unusually wide.

Anthropic states that memory import is available for free, Pro, Max, Team, and Enterprise users.

That removes one of the usual restrictions attached to migration features, since the tool is not locked behind a top-end plan.

A free user switching from another AI service can use the feature.

A Team or Enterprise user can also use it, subject to the memory controls that exist in those environments.

The surface map is more specific.

Anthropic explicitly confirms memory import on the web and Claude Desktop.

That wording should be kept exact.

It confirms those two surfaces directly and cleanly.

It does not, in the reviewed source set, clearly confirm that the same import flow is available on mobile.

This point should remain narrow in publication.

The broader Claude memory system may exist across more surfaces, and the release notes discuss memory more broadly, though the memory import article itself names web and Desktop for the import flow.

........

· Memory import is available across the full current Claude plan range.

· The confirmed import surfaces are the web app and Claude Desktop.

· Mobile import availability is not clearly confirmed in the reviewed official material.

........

Confirmed plan and surface coverage

Area	Confirmed status
Free	Supported
Pro	Supported
Max	Supported
Team	Supported
Enterprise	Supported
Web	Supported
Claude Desktop	Supported
Mobile import flow	Not clearly confirmed in reviewed sources

··········

IMPORT PIPELINE, INPUT STRUCTURE, AND HOW CLAUDE CONVERTS EXTERNAL MEMORY INTO ITS OWN FORMAT

The import flow is designed as a guided conversion pipeline in which the user supplies exported memory text and Claude converts it into reviewable internal edits rather than preserving the raw source as-is.

Anthropic documents a three-step process.

The user first exports memory from the previous AI service.

The user then opens Claude’s import flow, either through Settings > Capabilities > Memory > Start import or from the home-screen card labeled Import memory to Claude.

The user pastes the exported content into the text box and selects Add to memory.

Claude then extracts key information from that content and stores it as individual memory edits.

Anthropic also provides a recommended export prompt for the previous AI service.

That prompt asks the earlier assistant to list stored memory and learned context across categories such as response instructions, personal details, projects, goals, tools, languages, frameworks, and preferences or corrections, and to place the result in a single code block.

This is one of the more revealing parts of the feature.

Anthropic is not asking users to transport a proprietary memory file or a structured cross-vendor schema.

It is asking them to produce a readable memory summary and then pass that summary through Claude’s extraction layer.

Anthropic also explicitly notes that the prompt can be customized to remove sensitive information before import.

That makes the user’s own filtering step part of the intended design, not an afterthought.

........

· Anthropic documents a step-by-step import flow rather than a hidden or automatic migration process.

· The user supplies exported memory text, and Claude converts it into internal memory edits.

· Anthropic encourages users to customize the export prompt before import, especially to remove sensitive information.

........

Import pipeline from source assistant to Claude

Stage	What happens
Source export	The user exports memory from the previous AI service
Entry into Claude	The user opens Claude’s import tool from settings or the home-screen card
Transfer method	The user pastes exported content into the import box
Claude processing	Claude extracts key information from the pasted material
Final stored form	Individual memory edits inside Claude’s memory system

··········

RETENTION LOGIC, WORK-RELATED FILTERING, AND WHAT CLAUDE MAY LEAVE OUT

Imported memory is filtered through Claude’s own retention logic, which is explicitly oriented toward work-related usefulness rather than toward full preservation of every personal detail present in the imported source.

Anthropic states that Claude’s memory is designed to focus on work-related topics in order to improve collaboration.

That line defines the retention logic much more clearly than general product copy would.

Imported memory is therefore not treated as a neutral archive.

It is filtered through Claude’s own sense of what belongs inside its memory system.

Anthropic also states that Claude may not retain personal details unrelated to work.

This is one of the clearest scope boundaries in the entire feature.

A user may import a broader set of details from another assistant and still find that Claude does not preserve everything after processing.

This should not be described as an accidental edge case.

It is part of the documented design.

Anthropic also gives users a direct fallback when a specific item does not survive the import pass.

If there is information the user wants Claude to store, Anthropic directs them to Settings > Capabilities > View and edit your memory so that it can be added manually as a memory edit.

That shows how the system is meant to be used in practice.

The automated import provides the first pass.

The manual editing layer handles the corrections, additions, and missing details that the automated pass does not preserve.

........

· Claude’s memory is explicitly oriented toward work-related context.

· Personal details unrelated to work may not survive the import process.

· Manual memory editing is the documented fallback when important details are not retained automatically.

........

Retention logic after import

Type of imported content	Likely treatment inside Claude
Work-related preferences	Aligned with Claude’s memory design
Project and goal information	Strong fit for retention
Tools, languages, and frameworks	Strong fit for retention
Personal details unrelated to work	May not be retained
Missing but important details	Can be added manually through memory editing

··········

TIMING, REVIEW, AND HOW IMPORTED MEMORY BECOMES PART OF CLAUDE’S LARGER MEMORY FRAMEWORK

The result of an import is not only delayed by a documented processing window, but also folded into Claude’s wider memory system where it can be reviewed, toggled, and used alongside past-chat context.

Anthropic states that once the import is complete, updated memory should appear within 24 hours.

That timing window matters.

It means the effect is not guaranteed to be immediate in the same session or immediately after pasting the exported content.

The feature should therefore be understood as a memory update process inside Claude’s broader personalization framework rather than as an instant rewrite of runtime behavior.

Anthropic also states that the imported result can be reviewed through Manage edits.

This is an important control layer, especially because the feature is explicitly described as experimental.

The user is not expected to guess what Claude absorbed.

Claude is expected to surface the imported result as memory edits that can be examined.

The feature also sits inside the same broader memory system that can be toggled on or off in Settings > Capabilities.

Anthropic’s memory documentation further states that when Claude references previous conversations, users see citations back to the original chats.

So imported memory does not remain isolated in a migration-specific corner of the product.

It becomes part of Claude’s ongoing memory and prior-context framework.

That is what gives the feature practical long-term value after the initial transfer is complete.

........

· Imported memory should appear within 24 hours.

· The stored result can be reviewed through Manage edits.

· Imported memory feeds into Claude’s broader memory and prior-context system rather than remaining isolated.

··········

GOVERNANCE, ADMIN CONTROLS, AND THE CURRENT HARD LIMITS OF THE FEATURE

Claude import memory sits inside the same control architecture as Claude’s broader memory system, and the feature remains limited by experimental reliability, selective retention, and incomplete technical disclosure around import mechanics.

At the user level, Anthropic states that memory and past-chat search can be toggled on or off through Settings > Capabilities.

That means imported memory follows the same ordinary memory controls as the rest of Claude’s personalization layer.

For Enterprise, Anthropic adds organization-level controls.

Owners and Primary Owners can manage memory features through Organization settings > Capabilities.

Anthropic also states that disabling memory at the organization level immediately deletes existing memory synthesis data for all users in that organization, and individual users in that environment can no longer control the setting for themselves.

Anthropic further states that memory synthesis is encrypted at rest and tied to the underlying conversations, while chat summaries are stored alongside conversation data under existing retention policies.

The feature’s current hard limits are just as important.

Anthropic explicitly states that memory imports are experimental and still in active development.

It also states that Claude may not always successfully incorporate imported memories.

The reviewed official sources also do not publish a formal import schema, a maximum paste size, or a deterministic parsing specification for imported memory text.

This means the feature should not be presented as a standardized technical migration protocol.

It is better described as a broad user-facing import tool with real utility, wide plan access, and clear reliability and fidelity limits.

........

· Imported memory follows Claude’s existing user and enterprise memory controls.

· Enterprise admins can disable memory organization-wide, which deletes memory synthesis data for that organization.

· The feature is experimental, not fully deterministic, and not documented through a formal public import spec in the reviewed source set.

........

Governance and current feature limits

Area	Confirmed position
User control	Memory and past-chat search can be toggled on or off
Enterprise admin control	Owners and Primary Owners can manage memory features org-wide
Org-level disable effect	Existing memory synthesis data is deleted
Data handling	Memory synthesis is encrypted at rest
Reliability	Imported memories may not always be fully incorporated
Development state	Experimental and still in active development
Formal public import spec	Not clearly published in reviewed sources

·····

FOLLOW US FOR MORE.

·····

·····

DATA STUDIOS

·····

[datastudios.org]