Contents
  1. Purpose and foundations
    1. Memory across interactions
    2. Episodes and reusable claims
    3. Origins and turning points
  2. Forming accountable records
    1. Identity, scope, and provenance
    2. Choosing what to retain
    3. From observation to accepted memory
  3. Organizing and recalling memory
    1. Consolidation and independent support
    2. Records and searchable views
    3. Recall for the current task
  4. Use and change
    1. Personalization and authority
    2. Correction and supersession
    3. Expiry, deletion, and reappearance
  5. Evidence of useful memory
    1. Testing memory behavior
    2. Measuring the value of recall
    3. Frequently asked questions
      1. Does a longer context window give an agent cross-session memory?
      2. Does agent memory require a vector database?
      3. Can a deleted memory return?
  6. Check understanding
  7. Open questions
  8. Selected talks
  9. References
  10. Talk library
← All topics

Agent memory

Agent memory is information an application keeps from one interaction for use in a later decision, as Richmond Alake explains. In Shlok Khemani’s 2026 comparison, an early version of OpenAI’s ChatGPT extracted a dietary preference from what a user said and supplied it to later conversations. The user could inspect the saved item and delete it. Ben Holmes used a Markdown note he could correct directly.

Purpose and foundations

Memory across interactions

Follow the read path across these examples from the 2026 talks. Khemani studied particular versions of consumer assistants.
RepresentationWho writes it?What reaches a new session?Where can someone correct it?
Early ChatGPT saved factsThe assistant extracts an item from a user statement.Saved items are supplied to later conversations.The user could inspect the list and delete a mistaken item.
Later ChatGPT profile in Khemani’s testA background process synthesizes conversations.The assistant receives a derived account of the user.Khemani could identify a wrong assertion but could not inspect the raw profile in settings.
Early Claude conversation searchEarlier conversations supply the searchable messages.The assistant can look up past exchanges by topic or time.Khemani’s account of Anthropic’s implementation leaves the correction controls unspecified.
Holmes’s Markdown noteAuthor and enrichment skill.The file is available for lookup.In the file itself.

At 5:41 in Holmes’s demonstration, compare the dictated Ferrari note with its tags, source URL, enrichment timestamp, and backlinks. The skill can reuse existing tags. Follow a backlink to see which note refers to this one, or check a tag against Holmes’s original wording.

An agent is software that uses observations to choose actions. For a stored note to inform a model’s choice, some representation of it must enter the input to that call. A checkpoint has a different purpose: it preserves enough state to resume an execution. An inference cache saves numerical work performed while processing inputs. Neither specifies which past decisions a new conversation should inherit.

The distinction matters when an assistant seems to know something it learned yesterday. It might have received a saved item or a compressed account of the conversation. As Jack Morris explains, compression works on information already supplied. Training can instead change the model’s parameters so it behaves differently without receiving that record again. Editing Holmes’s file changes what a later reader can retrieve. It requires no such training. Continual Learning

Episodes and reusable claims

An episodic record describes what happened on a particular occasion. A transcript preserves the exchange in order. A shorter episode might keep just the request, the action, and the observed result. A semantic record expresses a claim for reuse across occasions. If someone asks for bullet points in this report, the episode tells us how they wanted that report formatted. Saving prefers bullet points would turn a one-time instruction into a default. If the user says use bullet points for future reports on this project, that wording supports a standing preference and tells the next session where it applies.

A semantic record can hold a stated preference, a checked fact, or a tentative inference. Its purpose is reuse. Semantic search is a method for finding records by similarity of meaning and works on either type. Sometimes the occasion is the better unit to retrieve. In Diane Lin’s alert-review example, a previous analyst judged a similar security alert benign. A new investigator can compare the circumstances even before anyone can state a general rule. If the precedent leaves the verdict disputed, an expert can identify the deciding condition. Keeping the original case makes that condition easier to check on the next alert.

Origins and turning points

Discussions of language understanding helped prompt Endel Tulving’s 1972 distinction between episodic and semantic memory. You can know what a word means without remembering when or where you heard it. In software the distinction describes what a record preserves. A system can look up a preference directly, but to explain how it learned that preference it needs a record of the occasion. Tulving’s retrospective

One early problem was how to learn enough about a person to help them without making every choice permanent. Elaine Rich’s 1979 GRUNDY book recommender retained individual reader profiles between sessions. At lookup, information specific to the dialogue took precedence over the persistent profile. A reader could ask for an unusual book without losing their usual preferences. Profile assertions also carried ratings and justifications. That made the reasoning behind a recommendation representable alongside the recommendation itself.

Another problem was finding an experience whose relevance became clear only later. Janet Kolodner’s 1983 paper on CYRUS describes answering questions about events involving Cyrus Vance and Edmund Muskie. A follow-up question could inherit clues from the previous exchange. Subject, time, and circumstances narrowed the events worth examining. In their 1994 case-based reasoning framework, Agnar Aamodt and Enric Plaza connected retrieval to testing. A system could reuse a previous solution, try it, revise it in light of the result, and select the useful parts for retention. Finding a precedent was the beginning of an investigation.

In the 2023 Generative Agents simulation, Joon Sung Park and colleagues stored observations and model-generated reflections that could inform later plans. The system could subsequently retrieve its own interpretation, making the distinction between observation and inference consequential. Charles Packer and colleagues’ MemGPT addressed a different constraint. The model could ask a function to search external storage, and the runtime would place a bounded page of results in the next input. The Letta workshop illustrates two write paths in this design. Conversation events populate searchable history automatically, while the agent deliberately deposits material in archival storage.

Forming accountable records

Identity, scope, and provenance

A message from a project lead saying that a reviewer wants email updates for this release describes the reviewer. An extractor that saves the preference under the lead’s personal profile has attached it to the wrong person. Even under the reviewer’s name, the claim needs its qualification: the lead reported it, and it applies to one release. Provenance records how the assertion came into existence, including the source, the transformation, and the actor responsible. It lets a reader inspect the basis for believing a claim. W3C provenance model

An application retaining the release preference could store the following distinctions. They give the maintainer a particular assertion to find, interpret, and correct.
Record detailMeaning in the example
Claim ID, revision, contentThe ID follows this assertion across edits. The revision identifies the wording a reader received.
Speaker, subject, recorderThe lead spoke about the reviewer. The extraction service wrote the record.
Attribution and evidenceMark the preference as reported by the lead. Link the specific message version. A stronger inference needs its own justification.
Scope, owner, readersLimit use to this release. Identify who controls the record and check who may read it.
Times and statusRecord when it applies, when the system learned it, and whether it remains current, disputed, superseded, or restricted.

What a source link says about a claim

Example

The original report and the evidence for revision have different roles.

Original source episodeContains the report from which the assertion was extracted.
Patient’s reported penicillin allergyA derived relationship whose source and validity can be inspected.
Later contrary episode, if presentRecords the evidence responsible for invalidation.
Graphiti can link a patient’s reported allergy to its source episode. The second path illustrates the general invalidation operation when contrary evidence arrives. It does not assert a clinical change in this patient.
Read the diagram as text
  • Original source episode. Contains the report from which the assertion was extracted.
  • Patient’s reported penicillin allergy. A derived relationship whose source and validity can be inspected.
  • Later contrary episode, if present. Records the evidence responsible for invalidation.
  • Original source episode → Patient’s reported penicillin allergy: source for extraction.
  • Later contrary episode, if present → Patient’s reported penicillin allergy: source for invalidation.

Use stable identifiers for the identities the system can establish. For an authenticated account, OpenID Connect specifies the combination of issuer and subject. An email address can change hands. A stable login still leaves two separate decisions: resolving a person mentioned by the speaker and authorizing the requester to read the resulting claim. Mem0’s entity-scoping documentation makes a related distinction between the speaker identifier on an extracted record and entities mentioned in its content. People and actors · Origins and dependencies

A knowledge graph represents entities and their relationships explicitly. In Daniel Chalef’s Graphiti demonstration for Zep, a relationship records a reported penicillin allergy. As Chalef puts it, “Tracing a fact to its source is just a graph walk.” Look at the roles of the links in the figure. One episode produced an assertion. A later episode can explain its invalidation. Counting both as votes for the assertion would reverse the meaning of the second link. If two entity records merge, the combined identity needs the source links from both. Recording at 4:54

Choosing what to retain

Memory formation starts with a decision about future use. What later question would this information answer? How quickly might it change? Who would bear the cost of a mistaken assumption? In an exploratory study involving six interviews and 54 public discussion threads, people described wanting writing preferences to carry between conversations while keeping health information out of academic writing. Some wanted overlap between projects under their control. A single account-wide rule would obscure precisely the distinction they cared about. Users’ Expectations and Practices with Agent Memory

Choose the smallest record that serves the identified purpose. These are example decisions an application owner could make.
Later needWhat to keepWhen to review or stop using it
Investigate a failed fixAn episode linking the attempted fix, observed result, and code version.Recheck the assumptions against current code before reusing the fix.
Format reports within a projectThe explicitly requested default, its project scope, and its source.Honor a local exception. Review the default when the project closes.
Determine what is deployed nowA reference to the authoritative deployment record.Read the live value for each current-status question. A dated snapshot serves a historical question.
No later task needs an incidental personal detailNo cross-session record.Check that automatic extraction also leaves the detail out of summaries.

An explicit request to remember gives evidence of intent for the stated use. Automatic extraction requires a policy about which observations qualify. Inferring a tendency introduces a further question about whether the evidence supports generalization. Across all three, a retention policy specifies purpose, readers, duration, review, and disposition, the action due at the end. A record can become useless before its permitted lifetime ends, or remain useful when deletion comes due. Tie the end action to an observable date or event. Delete the project default at closure gives a maintainer more direction than keep while useful. Artifact lifetimes · National Archives disposition guidance

From observation to accepted memory

Extraction proposes what a source means. Acceptance determines whether that proposal warrants changing a record. In Khemani’s account, conversations about choosing between Thailand and Turkey became a profile listing visits to both. He had traveled to Thailand and had never visited Turkey. The source supported considered Turkey. Saving visited Turkey changed the status of an event. A search that faithfully returns the saved profile will faithfully reproduce that mistake.

Use a direct mapping when an authoritative field already expresses the needed value and scope. For conversational text, compare the proposed record with its source. Check the speaker and subject, negation, time, and conditions. A plan needs evidence of completion before it becomes a completed event. For work performed through a tool, the tool result can supply that evidence where the agent’s own announcement cannot. If a permitted inference remains ambiguous and matters to a later decision, seek confirmation. Otherwise the application can retain the permitted episode without adopting the stronger claim.

The decision to write a reusable claim

Example

A permitted source can remain useful even when the proposed interpretation fails.

Source passageIncludes attribution and circumstances.
Proposed assertionIdentifies the subject and scope of the proposed change.
Check meaning and permitted useInspect the evidence and the record the proposal would change.
Commit the change
Seek confirmationKeep the assertion out of use while clarification is pending.
Leave the claim unwritten
An example admission policy compares a candidate with the source, retention rules, and existing state. A supported change can be committed. A material ambiguity can prompt clarification. An unsupported or prohibited assertion is omitted.
Read the diagram as text
  • Source passage. Includes attribution and circumstances.
  • Proposed assertion. Identifies the subject and scope of the proposed change.
  • Check meaning and permitted use. Inspect the evidence and the record the proposal would change.
  • Commit the change.
  • Seek confirmation. Keep the assertion out of use while clarification is pending.
  • Leave the claim unwritten.
  • Source passage → Proposed assertion: extraction.
  • Proposed assertion → Check meaning and permitted use: candidate and source link.
  • Check meaning and permitted use → Commit the change: supported and permitted.
  • Check meaning and permitted use → Seek confirmation: permitted, materially ambiguous.
  • Check meaning and permitted use → Leave the claim unwritten: unsupported or prohibited.

The Mem0 paper separates candidate extraction from updates. A new message pair, nearby messages, and a conversation summary supply candidates. A second stage retrieves related records and selects ADD, UPDATE, DELETE, or NOOP. This gives the application an identified change to inspect. A valid structured output tells the store how to interpret that operation. The evidence and retention policy determine whether to accept it.

A request-time write can make an accepted claim available before the application replies, at the cost of latency. A background writer can wait for more of the exchange, but the next session may arrive first. Be precise about acknowledgments. Receiving a message, committing an extracted claim, and publishing that claim to search are three observable events. An application that promises recall in the very next session needs a read path that can already see the committed revision.

Organizing and recalling memory

Consolidation and independent support

Consolidation reorganizes accumulated records for later use. It can merge duplicates or draw a lesson from several episodes. To judge that lesson, trace each claim back to its origin. In the figure, a test report says that reader A accepted an export format, and a project note quotes the result. Citing both documents still gives the agent one test. If the report is corrected, the quotation and any guidance based on it need review. The compatibility note about reader B has a different origin. It explains why the guidance for A should stay specific to A. Retrieving the guidance again adds no evidence for either claim.

Keep a derived lesson linked to its observations, qualifications, and conflicting evidence wherever source access and retention permit. A model-generated reflection interprets what happened. In Reflexion, an evaluator assesses an attempt and a model converts feedback into guidance for the next attempt. A mistaken explanation of a failure can persist even when the failure itself was observed correctly. Preserving the distinction allows a maintainer to withdraw the lesson while leaving an accurate episode intact.

Two documents can describe one test

Example

Following derivation links reveals which apparent sources share an origin.

Test report for reader AReader A accepted the tested export.
Excerpt in project notes
Compatibility note for reader BReader B requires a different format.
Reuse this format for reader AGuidance scoped to the tested reader.
In this example, an export test supports guidance for reader A. A project note repeats the test result. A separate compatibility note excludes reader B. Changing the test requires revisiting both support paths, while the limit for B has its own source.
Read the diagram as text
  • Test report for reader A. Reader A accepted the tested export.
  • Excerpt in project notes.
  • Compatibility note for reader B. Reader B requires a different format.
  • Reuse this format for reader A. Guidance scoped to the tested reader.
  • Test report for reader A → Excerpt in project notes: copied into.
  • Test report for reader A → Reuse this format for reader A: supports reuse for A.
  • Excerpt in project notes → Reuse this format for reader A: repeats the same test.
  • Compatibility note for reader B → Reuse this format for reader A: excludes B from scope.

Anthropic announced dreaming for Claude Managed Agents as a research preview on May 19, 2026. Its Dreams documentation describes consolidating a memory store and past session transcripts into a separate output store. The original stays available for comparison. An output ID can appear while work continues, and cancellation can leave partial results. Check completion and inspect the proposed changes before adopting the new store. By contrast, context compaction shortens the input used for ongoing work. A consolidated store can influence many future inputs.

The same traces can also become training material. Samuel Denton distinguishes offline training on accumulated production traces from improvement during live interaction. Ronak Malde describes a form of distillation, training a student from a teacher’s signal, in which the student generates attempts and the teacher receives additional hints. Later behavior can change because the student’s parameters changed. With an external store, the useful lesson must instead be made available to the later decision. That difference determines whether there is an individual record a user can inspect and edit.

Records and searchable views

Give the accepted record a stable identity and identify the revision a later task reads. In a compact profile, correcting one sentence can require reviewing other sentences derived from the same assumption. Individually addressable claims let the maintainer change one assertion and follow its dependencies. An episodic log preserves the order surrounding an event. A linked note preserves wording a reader can check directly. For instance, Polygraph’s session debrief can orient a coding task with a summary, then direct the investigator to the logs, pull request, and current code to check whether an old assumption still holds.

An index is a representation built to find records. In OpenClaw’s built-in memory engine, files hold the retained text and SQLite indexes it for keyword search. Optional embeddings represent passages numerically for similarity search. A full rebuild prepares a replacement database while queries continue using the published one. The engine publishes the replacement atomically after a successful rebuild. This preserves search availability even when a rebuild fails, but a direct file read can see an edit before search does.

A corrected record has two read paths

Example

Search can serve older wording after the authoritative write has succeeded.

Memory file, revision 2The correction is committed.
Authorized direct lookup
Published index with revision 1
Replacement index with revision 2Built separately from the serving index.
Search result for the same recordInspect the wording and revision returned.
A direct authorized lookup can see revision 2 while the published search index still contains revision 1. A successful rebuild replaces the published index. The two incoming paths to search represent successive index generations.
Read the diagram as text
  • Memory file, revision 2. The correction is committed.
  • Authorized direct lookup.
  • Published index with revision 1.
  • Replacement index with revision 2. Built separately from the serving index.
  • Search result for the same record. Inspect the wording and revision returned.
  • Memory file, revision 2 → Authorized direct lookup: current text by ID.
  • Memory file, revision 2 → Replacement index with revision 2: revised text for indexing.
  • Published index with revision 1 → Search result for the same record: old text until publication.
  • Replacement index with revision 2 → Search result for the same record: new text after publication.

Check the promised behavior at the point where a new session reads. Zep’s ingestion guide distinguishes accepted input, completed processing, and search readiness. Its search_when_ready helper can finish as soon as any result appears. That result could predate the edit. For a required correction, verify the intended ID, revision, and content through the actual search path, or supply the authorized current record by ID. Where record changes feed an external index, a transactional outbox can commit the change and the obligation to index it together. Search visibility still awaits delivery and publication.

Recall for the current task

Recall selects earlier evidence for the decision at hand. Supplying a small profile on every call makes common defaults available immediately, but irrelevant entries take up space and can distract the model. Looking up records on demand costs time and lets the agent choose how much detail to retrieve. A saved claim might give the chosen export format in one line. To explain why that format works, the agent may need the original test report. The claim and the report are different retrieval units. Either can narrow the next investigation if it answers the question the agent would otherwise have to research again. Yu Su on search and expertise

For a lookup of the reviewer’s update preference for this release, first decide which records may inform the task. Rank only eligible records by usefulness.
CandidateEligible?Consequence
Another reviewer’s almost identical requestNo. Wrong subject.Textual similarity cannot transfer a preference between people.
An earlier, superseded defaultNo for current use.It may qualify for a historical query.
A relevant report the requester cannot accessNo. Access restricted.Keep its contents out of the model input.
An accessible, current release preference phrased differentlyYes, within this release.Supply the preference with the reporter and qualification intact.

Use a known ID for exact lookup, date constraints for a past state, and similarity search for unfamiliar wording. Among eligible records, relevance, recency, and importance can help decide which to supply. Expand a compact claim to its supporting passage if the task turns on a missing qualifier. Check access on that expansion too. A scope filter enforces only its specified dimensions: Mem0 documents that a user-only filter can span applications and runs.

Deliver selected records with their versions, source roles, and applicable times to the context assembler. A completed lookup that finds no eligible record can justify asking for the missing information. An unavailable store or partially indexed import calls for a different explanation: the application could not complete the lookup. Treating either outcome as evidence that the user has no preference can turn a temporary read failure into another wrong memory. Degraded search coverage

Use and change

Personalization and authority

Personalization adapts behavior using information associated with a person or task. If someone has asked for bullet points in project updates, that standing preference can settle an open formatting choice. If today’s request specifies a narrative for the board, write the narrative. The standing preference can still apply to tomorrow’s ordinary update. An inferred tendency warrants a lighter touch. Repeated choices might all reflect the same temporary constraint, and silence gives little evidence about why someone accepted a result. Seek confirmation when the assumption would materially affect the outcome.

A correction is easier when the person can see both the claim and its basis. Cook and Kay’s 1994 Justified User Model exposed evidence for and against an inferred characteristic. A user edit added strongly weighted evidence from the user. The earlier justification remained inspectable. For a contemporary memory interface, this suggests a concrete distinction to expose when correcting a response: change the choice for this occasion, or change the default used on future occasions.

A source can be authoritative about one question and unable to settle another. A calendar can show that a time is free. Booking it requires current authorization. AuthMem-Bench tests this boundary by varying who supplied a claim while holding its content and the later task fixed. Stored text can also be hostile. Memory poisoning contaminates retained records so later retrieval can influence behavior. In AgentPoison, the attacker had access to an embedding model and could insert records. The attacker crafted malicious demonstrations that the retrieval system was likely to select when a later query contained a particular trigger. Familiarity with a retrieved instruction confers no permission to execute it. Current authority · AI Security

Correction and supersession

Before resolving a disagreement, check whether both claims describe the same subject, time, and circumstances. Neither arrival order nor confident wording can make that determination.
What the evidence showsHow to change the memory
The extraction misstated the original.Correct the assertion and review any summaries built from the error.
The old value was true until a real change.Supersede it from the effective date. Preserve its earlier interval where retention permits.
The new request applies to one occasion.Record or apply the narrower exception while keeping the broader default.
Two reports conflict about the same circumstances.Keep both attributed accounts available for investigation. Seek evidence with authority over the particular question.
An inference has insufficient support.Suspend its use. Ask for confirmation if the task depends on it.

In XTDB’s temporal model, valid time records when a claim applies in the world. Recorded time, called system time in XTDB, records when the database learned that version. Suppose a project finished on Monday, but its completion report arrived on Wednesday. Asked today about the project’s status at the end of Monday, the agent can use the report to answer that it was complete. Asked what evidence a decision maker had on Tuesday about that same Monday, the agent must exclude the report. At that cutoff, completion was still unreported. Graphiti records why a relationship was revised. In Chalef’s example, contradictory evidence prompts Graphiti to add an invalidAt date and link the relationship to the source episodes responsible for its invalidation.

A correction must reach claims derived from the old wording, saved summaries, and searchable copies. It can also race a writer that read the old revision. Require the writer to identify its starting revision and enforce that precondition when committing the edit. HTTP’s If-Match is a concrete example: a server compares the supplied strong validator with the current representation before performing the mutation. An intervening change invalidates the writer’s basis for overwriting the record. Reread the new version and reconsider the evidence before proposing another edit. Software Engineering Fundamentals

Accepting M revision 8 changes the authoritative record while an older summary remains available through ordinary recall. The summary names its source revision; this does not assign a universal revision scheme to indexes. Dashed work is pending, not completed propagation. Both paths are authorized in this logical example.

Expiry, deletion, and reappearance

A low-ranked record can still be retrieved. An expired preference may remain available to a historical investigation. A read restriction can exclude a record before its stored copies have been removed. To specify forgetting, identify both the end condition and the action it triggers. A time to live sets a duration according to a particular clock. In LangGraph’s store interface, expiry support depends on the adapter, reads normally refresh the timer when TTL is enabled, and deletion is best effort. A frequently read obsolete preference could therefore survive indefinitely under a sliding timer. A policy tied to project closure needs an explicit closure event.

Follow deletion from source to assertion. In Chalef’s example, an intake episode is the sole source for a contact preference. Removing it also removes that preference. A reported allergy survives because two other episodes still support it. Now consider the wording of a shared summary. A surviving source may establish the allergy but say nothing about how to contact the person. Zep’s deletion documentation says that surviving shared-node names and summaries are not regenerated. To preserve only the supported information, rebuild the text from the remaining permitted sources and check each surviving detail.

Forgetting while indexing is underway

Example

Cleanup can finish while a job still holds old text.

Source session
Buffered text
Forgotten session ID
Tracked copy removed
Publication check
At publication, the writer checks the buffered text’s source ID. A forgotten ID blocks the write.
Read the diagram as text
  • Source session.
  • Buffered text.
  • Forgotten session ID.
  • Tracked copy removed.
  • Publication check.
  • Source session → Buffered text: earlier read.
  • Source session → Forgotten session ID: forget request.
  • Forgotten session ID → Tracked copy removed: then clean up.
  • Buffered text → Publication check: proposed write.
  • Forgotten session ID → Publication check: exclusion.

OpenClaw removes the entire tracked entry if any merged source session is forgotten. The figure follows a job already holding that session’s text.

Report exclusion from future reads, completion of tracked cleanup, and residual copies separately. Check transcripts, exports, backups, and restore paths. If a copy must be retained for a restricted purpose, identify its owner and exclude it from ordinary reads and extraction. Reapply exclusions before a restored store can serve or publish. OpenAI’s Memory FAQ explains that retained older chats can generate memories again after memory is re-enabled. Deleting an application record also cannot retract an answer already delivered or change parameters trained on that information. Deletion propagation · Learned influence

Evidence of useful memory

Testing memory behavior

Build a test from an earlier interaction, the information permitted for retention, any intervening change, and a later task that does not give away the answer. Specify what should be available and what must be absent. Run independent cases against isolated histories. For a replay, freeze the information cutoff across originals, summaries, and indexes. Otherwise a summary can reveal a correction that arrived after the decision under test.

Hold the later task fixed and inspect the first boundary where the actual evidence differs from the expected evidence.
BoundaryDiscriminating test
FormationRead the committed record beside the original. Check subject, qualifiers, and whether a detail excluded by policy was written anywhere.
RecallInclude a near-match for another user and an expired claim. Compare permitted IDs and versions with what the model received.
CorrectionPause writes. If the source is current but search returns old wording under the right note ID, compare the saved note with its indexed copy. Old wording in the note locates a problem before indexing. Current wording in the note directs attention to publication or serving.
Evidence useSupply the correct permitted record directly. A wrong answer under direct delivery calls for inspecting interpretation, other inputs, and the task itself.
ForgettingPause a writer, forget its source, then resume it. Inspect the publication decision and stored derivatives. A clean answer alone cannot establish removal.

In an oracle condition, the evaluator supplies the record known to be relevant to each task. Stefania Druga compared this with no recall, similarity search, and lookup from a ledger of prior decisions while keeping the model fixed. Even with the correct record in its input, the model sometimes ignored or misused it. If ordinary recall fails where direct delivery succeeds, inspect the saved record first. If it misstates the source, investigate the write. If it is accurate, check which revision reached the model and how the model acted on it. A trace that captures these successive states lets the investigator follow the error. Observability

Trace the suspected transition before changing the policy. LangChain’s Vivek Trivedy asks whether deterioration first appears after the first or second compaction. Compare the inputs on either side. Did the relevant detail disappear, or did it remain visible while the action changed? Once a fix is proposed, Ben Hylak’s replay approach suggests testing past behavior for side effects. A change that restores one missing detail should also be tested on cases where that detail ought to stay out.

LoCoMo labels the turns containing answer evidence, allowing retrieval to be assessed separately from the answer. LongMemEval tests cross-session reasoning, knowledge updates, and abstention when the history lacks an answer. Memora specifies both facts a response should use and obsolete facts it must exclude. These constructed histories answer behavioral questions. Claims about user isolation, deletion completion, and stale writers require tests of the application’s actual storage and access paths.

Measuring the value of recall

A memory system earns its upkeep through changes in later work. In Parth Asawa’s database task, an agent repeatedly answers questions about a database. Once it has discovered which tables join, it might answer the next question with fewer exploratory SQL queries. Score correctness alongside query count. Guessing a join saves a query too. Run the same agent on matched questions with experience retained and with experience reset. Report absolute success in each condition as well as the gain attributable to carrying state forward.

Compare a richer memory policy with a credible cheaper alternative, such as recent history or a small explicit profile. For a component comparison, hold model, tasks, tools, and budgets fixed and verify that recall actually ran. For a comparison of complete systems, report actual resource use and the stopping rule. In Sara Hooker’s AutoScientist example, stopping once a target had been passed affected the distribution of observed win rates. A policy allowed more attempts has changed the experiment even if the memory store is identical.

Mean task success across five simulated task groups in MultiSessionCollab. The simulator withheld progress when a preference was violated. Each row compares the same model with and without retained preferences.
ModelWithout preference memoryWith preference memory
Llama-3.3-70B-Instruct41.78%46.38%
Qwen-2.5-7B36.21%35.18%

Llama’s users in the simulation also needed fewer preference reminders, down from 2.98 to 1.96 per session. Qwen’s success fell with memory enabled. In a separate study, 19 participants used assigned preferences over three sessions. Preference-related ratings improved, but participants described weaker transfer across domains, and some preferred to restate their choices. These conditions support a bounded comparison. They leave open how much time a person would save after months of maintaining and correcting a profile.

For a live comparison, assign users or workspaces to separate policies and isolate their histories. Switching policies per request would let one inherit the other’s writes. Separate tasks that depend on earlier experience from tasks where carryover should have no effect. Count correct continuations and repeated clarification alongside stale assumptions, inappropriate disclosures, and correction effort. Report read latency, write latency, and consolidation cost in their own units. Treat missing feedback as unknown: a user who stops correcting a system may have stopped relying on it. Evals and Benchmarks

Frequently asked questions

Does a longer context window give an agent cross-session memory?

A longer window increases the amount of information that fits in one model call. To use a decision from another session, the application must retain it and supply it again, either automatically or after a lookup. MemGPT’s paginated search is one way to bring just the relevant part of external history into the next call.

Does agent memory require a vector database?

No. A keyed record can supply a known preference, and a file can preserve a note for later reading. OpenClaw supports keyword search over retained files using SQLite. Vector search adds a way to find passages expressed in different words.

Can a deleted memory return?

Yes. A retained chat can supply the information again during extraction. A queued job can also recreate a deleted entry from text it read earlier.

Open questions

  1. When does repeated behavior justify a standing preference? Several choices may share one temporary constraint. Progress would mean predicting which defaults transfer to new tasks while preserving exceptions and reducing the work users spend correcting them.

  2. How much of a synthesis can survive selective forgetting? A remaining source can support the main claim while a single phrase exposes the removed source. A convincing result would regenerate the useful text from permitted evidence, verify the served revision, and withstand a delayed writer holding the old text.

  3. Which policies save effort over months of real use? As people come to expect continuity, they change what they explain and which failures they report. A field comparison with isolated histories would need to measure correct continuation, inappropriate reuse, correction time, and operating cost together.

Follow the curated reading path through the speakers and demonstrations behind this entry.

Explore more talks

The rest of the library, beyond the curated path. Cited talks support this entry; reviewed transcripts were processed in full. Metadata candidates have not been reviewed as sources or verified as topic members.

156 matching talks

TalkSpeakerEventYear
Richmond AlakeAI Engineer World's Fair 20252025
Charles PackerAI Engineer Summit 20252025
Stop Using RAG as Memory

Transcript reviewed

Daniel ChalefAI Engineer World's Fair 20252025
Stefania DrugaAI Engineer World's Fair 20262026
Ilan BigioAI Engineer Summit 20252025
Kam LasaterAI Engineer Summit 20252025
Greg BensonAI Engineer World's Fair 20252025
Divakar KumarAI Engineer World's Fair 20262026
Michael Hunger, Stephen Chin, Jesús BarrasaAI Engineer World's Fair 20252025
Philipp KrennAI Engineer World's Fair 20252025
David KaramAI Engineer World's Fair 20252025
Stephen ChinAI Engineer Europe 20262026
Louis-François Bouchard, Omar Solano, Samridhi VaidAI Engineer World's Fair 20262026
Dex HorthyAI Engineer Code 20252025
Sally-Ann DeLuciaAI Engineer Europe 20262026
Eric AllamAI Engineer World's Fair 20252025
Vinoth GovindarajanAI Engineer World's Fair 20262026
Harshil AgrawalAI Engineer Europe 20262026
Lovina DmelloAI Engineer World's Fair 20262026
Nick HeinerAI Engineer World's Fair 20262026
Tengyu MaAI Engineer World's Fair 20252025
Soheil FeiziAI Engineer World's Fair 20262026
Ishita DagaAI Engineer World's Fair 20262026
Paul Iusztin, Louis-François BouchardAI Engineer World's Fair 20262026
Diane LinAI Engineer World's Fair 20262026
A Genius With Amnesia

Transcript reviewed

Victor SavkinAI Engineer World's Fair 20262026
Dr. Sajjan KanukolanuAI Engineer World's Fair 20262026
Vinoo GaneshAI Engineer World's Fair 20262026
Building security around ML

Transcript reviewed

Dr. Andrew DavisAI Engineer World's Fair 20242024
Charles FryeAI Engineer Summit 20232023
Kuba RogutAI Engineer Europe 20262026
Kuba RogutAI Engineer Europe 20262026
Raj NavakotiAI Engineer Europe 20262026
Yohei NakajimaAI Engineer World's Fair 20262026
Christopher Lovejoy, Saul HowardAI Engineer World's Fair 20262026
Lance MartinAI Engineer World's Fair 20262026
Brandon WaselnukAI Engineer Europe 20262026
Sai Krishna RallabandiAI Engineer World's Fair 20262026
Shivam VermaAI Engineer World's Fair 20252025
Build a Prompt Learning Loop

Transcript reviewed

SallyAnn DeLucia, Fuad AliAI Engineer Code 20252025
Alberto RomeroAI Engineer Code 20252025
Šimon PodhajskýAI Engineer Europe 20262026
Ahmed MenshawyAI Engineer World's Fair 20242024
Garrett GalowAI Engineer Europe 20262026
Aparna Dhinkaran, Aparna DhinakaranAI Engineer Summit 20252025
Anton TroynikovAI Engineer Summit 20232023
Ofer MendelevitchAI Engineer Summit 20252025
Proactive Agents

Transcript reviewed

Kath KorevecAI Engineer Code 20252025
Scaling Compute on Context

Cited in this entry

Jack MorrisAI Engineer World's Fair 20262026
Yu SuAI Engineer World's Fair 20262026
Samuel DentonAI Engineer World's Fair 20262026
Ronak MaldeAI Engineer World's Fair 20262026
Sara HookerAI Engineer World's Fair 20262026
Ben HylakAI Engineer World's Fair 20262026
Stephen ChinAI Engineer World's Fair 20262026
Mark Bain, Vasilije Markovic, Daniel Chalef, Alex GilmoreAI Engineer World's Fair 20252025
Josh PurtellAI Engineer World's Fair 20252025
James LeAI Engineer World's Fair 20262026
Jack MorrisAI Engineer Code 20252025
Hubert MisztelaAI Engineer World's Fair 20252025
Stephen ChinAI Engineer World's Fair 20252025
Leonie MonigattiAI Engineer Europe 20262026
Agents Building Agents

Metadata candidate

Alfonso GrazianoAI Engineer World's Fair 20262026
Gabe De MesaAI Engineer World's Fair 20262026
Agents Need Feature Flags

Metadata candidate

Sachin GuptaAI Engineer World's Fair 20262026
Nick Nisi, Lizzie SiegleAI Engineer World's Fair 20252025
Anita KirkovskaAI Engineer Summit 20252025
Matt PocockAI Engineer Europe 20262026
Charles FryeAI Engineer Summit 20232023
AI SDK v6

Metadata candidate

Nico AlbaneseAI Engineer Europe 20262026
Michal CichraAI Engineer Europe 20262026
Paul Klein IVAI Engineer World's Fair 20262026
Build Systems, Not Code

Metadata candidate

Angie JonesAI Engineer World's Fair 20262026
Mahesh MuragAI Engineer Summit 20252025
Nishant GuptaAI Engineer World's Fair 20262026
Apoorva JoshiAI Engineer World's Fair 20252025
Michael FesterAI Engineer World's Fair 20252025
Building Self-Coding Agents

Metadata candidate

Colin FlahertyAI Engineer Summit 20252025
Liam McGarrigleAI Engineer Europe 20262026
Cat Wu, Thariq Shihipar, Simon WillisonAI Engineer World's Fair 20262026
Morgante PellAI Engineer World's Fair 20242024
Sunil PaiAI Engineer Europe 20262026
Stephen ChinAI Engineer Code 20252025
Andreas Kollegger, Zaid ZaimAI Engineer Europe 20262026
Val Bercovici, Callan FoxAI Engineer Code 20252025
Convex Launch

Metadata candidate

Jamie TurnerAI Engineer World's Fair 20242024
Develop at Idea Velocity

Metadata candidate

Jeffrey Lee-ChanAI Engineer World's Fair 20262026
Abi AryanAI Engineer Summit 20232023
Ara KhanAI Engineer Europe 20262026
Sam BhagwatAI Engineer World's Fair 20262026
Katelyn LesseAI Engineer Code 20252025
Sarah ChiengAI Engineer Europe 20262026
Joel HronAI Engineer World's Fair 20252025
Jason LiuAI Engineer World's Fair 20262026
Florina Muntenescu, Oli GaymondAI Engineer Europe 20262026
Phoebe KlettAI Engineer World's Fair 20242024
Jonathan LarsonAI Engineer World's Fair 20252025
Mithun HunsurAI Engineer Summit 20232023
Vasant KearneyAI Engineer World's Fair 20262026
Kyle Jaejun LeeAI Engineer World's Fair 20262026
Mahmoud AbdelwahabAI Engineer Code 20252025
Chip HuyenAI Engineer Summit 20252025
Raymond FengAI Engineer World's Fair 20262026
Xiaofeng WangAI Engineer Summit 20252025
Rukma SenAI Engineer World's Fair 20242024
Matthias LoiblAI Engineer World's Fair 20252025
Cornelia DavisAI Engineer World's Fair 20262026
Minimax M2

Metadata candidate

Olive SongAI Engineer Code 20252025
Lech KalinowskiAI Engineer World's Fair 20262026
Recursive Coding Agents

Metadata candidate

Raymond WeitekampAI Engineer World's Fair 20262026
Max RyabininAI Engineer Europe 20262026
Mozhgan Kabiri ChimehAI Engineer Europe 20262026
Merve NoyanAI Engineer Europe 20262026
Sarah GuoAI Engineer World's Fair 20252025
Karan GoelAI Engineer World's Fair 20242024
Apoorva Joshi, Ben PerlmutterAI Engineer World's Fair 20242024
Jack CableAI Engineer World's Fair 20262026
Dylan PatelAI Engineer World's Fair 20252025
Cormac BrickAI Engineer Europe 20262026
Sonam PankajAI Engineer World's Fair 20262026
João MouraAI Engineer World's Fair 20242024
Rajkumar SakthivelAI Engineer World's Fair 20262026
What the Best Agents Share

Metadata candidate

Mardu SwanepoelAI Engineer Europe 20262026
Ahmad AwaisAI Engineer World's Fair 20252025
Zach BlumenfeldAI Engineer Europe 20262026
Eugene CheahAI Engineer Summit 20252025
Ben BurtenshawAI Engineer Europe 20262026
Erina KaratiAI Engineer World's Fair 20262026
Noyan TokgozogluEvent unknownUnknown
Howie Liu, Shawn "swyx" Wang, Joshua XuAI Engineer World's Fair 20262026
Jeremy AdamsEvent unknownUnknown
Jordi SprangerAI Engineer NYC 2026 Online Talks2026
Kay MalcolmEvent unknownUnknown
Dylan CouzonAI Engineer Code 2025Unknown
Suraj GuptaAI Engineer World's Fair 20262026
Zack Proser, Nick NisiEvent unknownUnknown
Aparna DhinakaranEvent unknownUnknown
Charlie Guo, Gabriel ChuaAI Engineer World's Fair 20262026
Paula DozsaEvent unknownUnknown
Jake BroekhuizenAI Engineer World's Fair 20262026
Priyanka Phatak, Gabe CemajAI Engineer World's Fair 20262026
Joseph KimEvent unknownUnknown
Harish GaggarAI Engineer NYC 2026 Online Talks2026
William LyonEvent unknownUnknown
Andrey RybkaEvent unknownUnknown
Mike ChambersEvent unknownUnknown
Anders SwansonAI Engineer World's Fair 20262026
Ignacio MartinezAI Engineer World's Fair 20262026
Jediah KatzEvent unknownUnknown
Fuad AliEvent unknownUnknown
Gabe Cemaj, Priyanka PhatakAI Engineer World's Fair 20262026
Sam Green, Aaron LinskyEvent unknownUnknown
Aman SinghEvent unknownUnknown
Karthik Ranganathan, Heather DowningAI Engineer World's Fair 20262026
Nacho MartinezEvent unknownUnknown
Elizabeth Fuentes LeoneAI Engineer World's Fair 20262026

References

Coverage and source review
Processed transcripts
51 processed in full · 5 in the curated path
Automated source review
Passed
Metadata candidates
102 unreviewed; not verified topic membership
Corpus version
1bd8e407b26a07b33815594e1b2db5f41827119a2b3cb6fbf240f9fc571fc767

Automated review checks source support; it is not publication approval.

A synthesis of selected conference talks and technical references. Citations link to the source material; they do not imply that every talk on this subject is included.

  1. Architecting Agent Memory: Principles, Patterns, and Best Practices — Richmond Alake

    https://www.youtube.com/watch?v=W2HVdB4Jbjs&t=353s

  2. Lessons from Studying Every Memory System

    https://www.youtube.com/watch?v=5ZGyKWjQDr0&t=140s

  3. Lessons from Studying Every Memory System

    https://www.youtube.com/watch?v=5ZGyKWjQDr0&t=356

  4. Lessons from Studying Every Memory System

    https://www.youtube.com/watch?v=5ZGyKWjQDr0&t=378s

  5. LLM Knowledge Bases: a practical guide

    https://www.youtube.com/watch?v=I3bpdgFJCUY&t=341s

  6. MemGPT: Towards LLMs as Operating Systems

    https://arxiv.org/html/2310.08560v2

  7. Breaking the Chain: Agent Continuations for Resumable AI Workflows

    https://www.youtube.com/watch?v=ZB7l4uxW3Yo&t=144

  8. Scaling Compute on Context

    https://www.youtube.com/watch?v=WiqDvX6isc4&t=769s

  9. Stateful Agents — Full Workshop with Charles Packer of Letta and MemGPT

    https://www.youtube.com/watch?v=E0k9Ppq6yXY&t=170

  10. Cognitive Architectures for Language Agents

    https://arxiv.org/html/2309.02427v3

  11. When Memory Becomes Authority: Benchmarking Authority Collapse at the Memory Consolidation Boundary

    https://arxiv.org/html/2608.01679v1

  12. Memory overview — LangChain

    https://docs.langchain.com/oss/python/concepts/memory

  13. Why Your Agent Disagrees With Itself (And What To Do About It)

    https://www.youtube.com/watch?v=wEc9aG7cRQc&t=1089

  14. Why Your Agent Disagrees With Itself (And What To Do About It)

    https://www.youtube.com/watch?v=wEc9aG7cRQc&t=1203

  15. Endel Tulving: Citation Classic commentary on Episodic and Semantic Memory

    https://garfield.library.upenn.edu/classics1987/A1987K743800001.pdf

  16. User Modeling via Stereotypes — Elaine Rich

    https://www.cs.utexas.edu/~ear/CogSci.pdf

  17. Reconstructive Memory: A Computer Model

    https://folk.idi.ntnu.no/agnar/CBR%20papers/kolodner83reconstructive.pdf

  18. Case-Based Reasoning: Foundational Issues, Methodological Variations, and System Approaches

    https://www.iiia.csic.es/~enric/papers/Aamodt_1994_Case.pdf

  19. Generative Agents: Interactive Simulacra of Human Behavior

    https://arxiv.org/html/2304.03442v2

  20. MemGPT: Towards LLMs as Operating Systems

    https://arxiv.org/html/2310.08560v2

  21. Stateful Agents — Full Workshop with Charles Packer of Letta and MemGPT

    https://www.youtube.com/watch?v=E0k9Ppq6yXY&t=2032

  22. PROV-O: The PROV Ontology — W3C

    https://www.w3.org/TR/prov-o/

  23. Entity-Scoped Memory — Mem0

    https://docs.mem0.ai/platform/features/entity-scoped-memory

  24. PROV-DM: The PROV Data Model

    https://www.w3.org/TR/prov-dm/

  25. Time in XTDB

    https://docs.xtdb.com/about/time-in-xtdb.html

  26. OpenID Connect Core 1.0 incorporating errata set 2

    https://openid.net/specs/openid-connect-core-1_0.html

  27. Citation Needed: Provenance for LLM-Built Knowledge Graphs — Daniel Chalef

    https://www.youtube.com/watch?v=H7puB0RwJMM&t=294s

  28. Citation Needed: Provenance for LLM-Built Knowledge Graphs

    https://www.youtube.com/watch?v=H7puB0RwJMM&t=316

  29. Episode metadata projection — Zep

    https://help.getzep.com/v3/episode-metadata-projection

  30. Users' Expectations and Practices with Agent Memory

    https://brennanjones.com/media/documents/publications/chiea25-666.pdf

  31. Institutional memory — Polygraph

    https://docs.trypolygraph.com/features/institutional-memory/

  32. Preparing Disposition Instructions — National Archives

    https://www.archives.gov/records-mgmt/scheduling/instructions

  33. Stop babysitting your agents: building a context engine for mergeable code

    https://www.youtube.com/watch?v=BiG2ssibKGc&t=785

  34. Custom Instructions — Mem0

    https://docs.mem0.ai/platform/features/custom-instructions

  35. Active Graph Agent Runtime (BabyAGI 4)

    https://www.youtube.com/watch?v=khVX_BUnEwU&t=822

  36. Mem0: Building Production-Ready AI Agents with Scalable Long-Term Memory

    https://arxiv.org/html/2504.19413v1

  37. Check Data Ingestion Status — Zep

    https://help.getzep.com/check-data-ingestion-status

  38. Reflexion: Language Agents with Verbal Reinforcement Learning

    https://arxiv.org/html/2303.11366v4

  39. New in Claude Managed Agents: dreaming, outcomes, and multiagent orchestration

    https://claude.com/blog/new-in-claude-managed-agents

  40. Dreams — Claude Managed Agents

    https://platform.claude.com/docs/en/managed-agents/dreams

  41. How We Solved Context Management in Agents — Sally-Ann DeLucia

    https://www.youtube.com/watch?v=esY99nYXxR4&t=374

  42. Bringing Continual Learning into Enterprises

    https://www.youtube.com/watch?v=ZTA0GwpAUak&t=170s

  43. Scaling up Continual Learning

    https://www.youtube.com/watch?v=zL1kLftVTlo&t=449s

  44. Memory overview — LangChain

    https://docs.langchain.com/oss/python/concepts/memory

  45. Active Graph Agent Runtime (BabyAGI 4)

    https://www.youtube.com/watch?v=khVX_BUnEwU&t=501

  46. Built-in Memory Engine — OpenClaw

    https://docs.openclaw.ai/concepts/memory-builtin

  47. Transactional outbox pattern — AWS Prescriptive Guidance

    https://docs.aws.amazon.com/prescriptive-guidance/latest/cloud-design-patterns/transactional-outbox.html

  48. Intelligence + Continual Learning = Expertise

    https://www.youtube.com/watch?v=I6aiEf3aEFQ&t=588s

  49. Memory Harnesses for Long-Running Research Agents

    https://www.youtube.com/watch?v=R3-anFK1YM8&t=602

  50. Stateful Agents — Full Workshop with Charles Packer of Letta and MemGPT

    https://www.youtube.com/watch?v=E0k9Ppq6yXY&t=386

  51. Citation Needed: Provenance for LLM-Built Knowledge Graphs

    https://www.youtube.com/watch?v=H7puB0RwJMM&t=294

  52. The Justified User Model: A Viewable, Explained User Model

    https://www.researchgate.net/publication/2820142_The_Justified_User_Model_A_Viewable_Explained_User_Model

  53. AgentPoison: Red-teaming LLM Agents via Poisoning Memory or Knowledge Bases

    https://arxiv.org/html/2407.12784v1

  54. Stop babysitting your agents: building a context engine for mergeable code

    https://www.youtube.com/watch?v=BiG2ssibKGc&t=558

  55. RFC 9110: HTTP Semantics — If-Match

    https://www.rfc-editor.org/rfc/rfc9110.html#section-13.1.1

  56. LangGraph BaseStore: updates, deletion and TTL

    https://raw.githubusercontent.com/langchain-ai/langgraph/main/libs/checkpoint/langgraph/store/base/__init__.py

  57. Citation Needed: Provenance for LLM-Built Knowledge Graphs

    https://www.youtube.com/watch?v=H7puB0RwJMM&t=587

  58. Deleting Data from the Graph — Zep

    https://help.getzep.com/deleting-data-from-the-graph

  59. Memory Provenance and Forgetting — OpenClaw

    https://docs.openclaw.ai/concepts/memory-provenance

  60. Memory FAQ — OpenAI

    https://help.openai.com/en/articles/8590148-memory-faq

  61. Elasticsearch: Delete a document

    https://www.elastic.co/docs/api/doc/elasticsearch/operation/operation-delete

  62. From Recall to Forgetting: Benchmarking Long-Term Memory for Personalized Agents

    https://arxiv.org/html/2604.20006v1

  63. Time in XTDB

    https://docs.xtdb.com/about/time-in-xtdb.html

  64. Evaluating Very Long-Term Conversational Memory of LLM Agents

    https://aclanthology.org/2024.acl-long.747.pdf

  65. Memory Harnesses for Long-Running Research Agents

    https://www.youtube.com/watch?v=R3-anFK1YM8&t=518

  66. Memory Harnesses for Long-Running Research Agents

    https://www.youtube.com/watch?v=R3-anFK1YM8&t=268

  67. OpenTelemetry GenAI semantic conventions: spans

    https://github.com/open-telemetry/semantic-conventions-genai/blob/main/docs/gen-ai/gen-ai-spans.md

  68. Improving Agents is a Data Mining Problem

    https://www.youtube.com/watch?v=CvRngaQZQ3Y&t=281s

  69. Designing Agents (The Floor Is the Frontier)

    https://www.youtube.com/watch?v=jHMiYtjoJfA&t=334s

  70. Context Engineering in 2026: Compaction, Memory & Cost

    https://www.youtube.com/watch?v=WP3hjUXd918&t=2509

  71. LongMemEval: Benchmarking Chat Assistants on Long-Term Interactive Memory

    https://arxiv.org/html/2410.10813v2

  72. Beyond Static Intelligence: Evaluating Continual Learning

    https://www.youtube.com/watch?v=iqloyWCGYQQ&t=639s

  73. Beyond Static Intelligence: Evaluating Continual Learning

    https://www.youtube.com/watch?v=iqloyWCGYQQ&t=494

  74. Context Engineering in 2026: Compaction, Memory & Cost

    https://www.youtube.com/watch?v=WP3hjUXd918&t=2584

  75. Adaption Labs — Gradient-Free Continual Learning

    https://www.youtube.com/watch?v=XEd_SRVHBgU&t=397s

  76. Learning User Preferences Through Interaction for Long-Term Collaboration

    https://arxiv.org/html/2601.02702v1

  77. Claude for long-horizon tasks

    https://www.youtube.com/watch?v=9QebvrrY3KY&t=1473

  78. Conversation state: managing the context window — OpenAI

    https://developers.openai.com/api/docs/guides/conversation-state

  79. Event time and actual availability require separate handling

    https://github.com/feast-dev/feast/issues/6615