Vertical / domain-specific AI memory

Clinician-assist ambient documentation. Source mapping: every AI-generated summary element traced back to the source utterance. Audit-and-trust layer over episodic memory. Built on proprietary 1.5M+ medical-encounter dataset.

Treats memory as first-class architecture. Captures the digital footprint of every interaction; retrieves preference and history at engagement time. Public example: airline knowing a frequent flyer wants aisle seats with her son — prefer…

Target identification / drug repurposing / mechanism tracing. 85+ data sources, petabyte-scale, rebuilt every few weeks. Wet-lab results re-enter the graph and shift downstream predictions — institutional experimental memory closing a fe…

Drug discovery / target identification / causal mechanism tracing. The graph is the memory: 7 years of curated biomedical cause-effect relationships compounding with each new ingestion. Scientific RAG retrieves from a versioned causal su…

Chat Memories (user-defined facts), auto-memories for c.ai+ subscribers, pinned memories, in-context retention. PipSqueak 2 model (April 2026) reduces in-conversation drift. Memory Visualization meter forthcoming.

No-code interactive narrative platform. Characters track what they know, when they were lied to, what they misremember; story state changes downstream behaviour. Characters can deliberately lie / misremember and track when players have l…

Enterprise contact centre (voice + chat). Short-term covers current session; long-term persists across sessions and is shared by every agent a customer might encounter. Most explicitly engineered memory model in the segment.

Mimir hierarchical memory: short-term (verbatim), medium-term (summaries + emotion), long-term (cross-session). Scene awareness layer adds in-engine perception. Unity + Unreal SDKs with real-time voice + lipsync.

Autonomous support agents resolving end-to-end (refunds, cancellations). Channel-agnostic memory: chat → call carries prior context. Agents recognise returning customers and recall prior issues.

Specialty-care ambient scribe (oncology focus). Pulls forward prior notes with precise dates; synthesises interval history (what changed since last visit); auto-populates disease-status and treatment-plan context.

Adds GPT-4-powered Roleplay (character conversations, with cross-call memory for "Lily") and Explain My Answer on top of Duolingo's existing spaced-repetition engine, which tracks per-word and per-skill mastery state across all sessions.

Persistent pronunciation profile per user, tracking which sounds, words, and patterns the learner consistently mispronounces. Self-evolving model re-evaluates and updates the profile as the learner improves. Competitive moat is a proprie…

Plaintiff litigation. Shared Firm Knowledge (firm-wide institutional store) + autonomous Auditor agent that runs nightly across all open matters surfacing missed value drivers (TBIs, unfiled MRI orders, mass-tort eligibility) without pro…

System 2 (7B VLM) feeds tokens to System 1 (80M visuomotor policy at 200 Hz). Temporal memory module maintains visual-state history; force-feedback buffered as tactile memory proxy. 500 hours of human demos. Deployed in BMW logistics.

Embedded in case-management platform. Holds full scope of every matter — deadlines, billing, discovery, communications, negotiation history — as live context for every query. Case data is the context, not a separate retrieval corpus.

Full autonomous-scientist loop. Owl agent surfaces prior work so downstream agents don't re-derive known results. WikiCrow produced cited factsheets for 15,616 previously undocumented human genes; Phoenix uses accumulated experiment cont…

Personal lawyer memory + matter memory + institutional firm memory + client-institution memory. Opt-in at user level with admin kill-switches and ethical-wall controls. Announced Jan 2026.

Patient-conversation / chronic-care staffing model. HIPAA-compliant Memory Store holds non-EHR salient facts from prior calls, persists across engagements, explicitly not used for model training. Shifts from transactional call-handling t…

Leading purpose-built NPC engine. Long-Term Memory (named feature) stores interaction facts across sessions; surfaces them contextually. Character Brain coordinates multiple ML models. Personality defined by trait axes, backstory, goals,…

Primary care / virtual physician intake. Two-agent pipeline: one extracts the relevant EMR subset, second answers grounded to that filtered history. Architecture explicitly diverges by patient condition (diabetic vs heart-failure get dif…

Therapeutic-antibody discovery in a closed Design-Build-Test-Learn loop. EVA accumulates structured outcome data from each robotic experiment cycle and feeds it back to generative ML models directing the next design round. Genuine experi…

AI Dungeon's maker. Voyage (April 2026) launched as a creator platform for AI RPGs. World Engine (5 years in dev) is dedicated AI tracking characters, objects, backstories, relationships independently from the narrative AI; persistent wo…

Jan 2026 platform relaunch. Persists negotiation history, prior decisions, accepted risk positions across all contracts; new terms align with what the org has already agreed. Branded as the antidote to "enterprise amnesia."

Teacher productivity platform (80+ teacher tools, 50+ student tools) for planning, differentiation, IEP generation, feedback. Does not maintain a cross-session learner model. Teachers input student context manually each session; district…

Ambient scribe + longitudinal clinical intelligence (strategic partnership Jul 2025). Live encounter transcript reconciled against historical labs/imaging/notes; condition-suspecting + care-gap engine runs against the reconciled record.

Tracks per-card answer history within a study set; computes a Memory Score for SRS scheduling. Magic Notes converts uploaded content (notes, PDFs, URLs) into flashcard sets automatically. Q-Chat, the conversational AI layer, was disconti…

Mirrors human cognitive functions: explicit memory, vision, hearing, tactile. Translates natural-language instructions into actions with explainable intermediate reasoning. Memory stores task procedures so learned skills are reusable acr…

Bret Taylor + Clay Bavor. Agent Data Platform links unstructured conversation data to enterprise systems (billing, inventory, policies, transactions); persists across channels and time. Memory accumulates across every interaction, not ju…

Real-time simultaneous inference + data collection — dynamically updates in-context representation as the robot encounters new situations. Lets the system adapt to novel embodiments / unseen environments without explicit fine-tuning. Pre…

Transactional / contract drafting. Ingests entire precedent corpus (OneDrive, Dropbox); ML surfaces and reuses the firm's own prior clauses during live drafting in Word, tuned to specific deal context. Goal: eliminate "reads like ChatGPT…

Doctor-assist ambient documentation. Pulls forward prior documentation into a pre-visit synthesis rather than waiting for the encounter. Patient summaries surface relevant insight from history before visit begins.

Single end-to-end neural network ingesting multi-camera video + navigation + ego-motion. Maintains evolving 3D occupancy volume (voxel-level) + 3D Gaussian rendering as internal world state — spatial memory persisting across frames; trac…

All surround-camera views encoded through video tokeniser to continuous latent; past latent sequences serve as explicit temporal context via space-time factorised transformer. GAIA-3 (15B params, 2× GAIA-2 size, 10× training data) adds s…

Extends π0 with MEM layer giving the policy short-term (within-task) + long-term (cross-task, >10-min horizon) memory. Co-trained on robot teleoperation + human video + text. Hierarchical inference: VLA predicts subtasks as text tokens, …

SMB law firms. AI embedded in practice management; pulls full matter overviews — WIP, billing, documents, communications — as context. Memory is the PMS data; no separately persisted cross-matter layer. Renamed Manage AI at ClioCon 2025.

Aug 2025 relaunch added agentic deep-research workflows (plan-execute-cite). Holds context within a session; no persistent cross-session matter memory layer publicly documented yet. Thomson is building proprietary "Thomson" legal LLM for…

Sales + contact centre. Customer history feeds real-time guidance and coaching. Memory survives handoffs between AI agent, human agent, and channels — but leans on CRM as the primary store.

Systematic literature review + data extraction + meta-analysis. Every extracted claim anchored to its exact source sentence; Notebook persists paper collections across projects. Research Agents (Dec 2025) carry evidence forward across re…

Corporate / transactional due diligence. Parallel AI agents across large document sets with firm-specific configurations and review workflows. Configuration amounts to a shallow institutional memory layer; cross-matter persistent memory …

Ambient documentation with clinician-profile learning. System learns each clinician's preferred note structure, phrasing patterns, symptom-outlining conventions — applies persistently across sessions. Provider-side personalisation rather…

CX support. Currently CRM-backed customer attributes + conversation history lookup. Lifecycle-persistent memory ("priorities learned day one considered in year 10") on roadmap as of Pioneer 2025 — not yet shipped.

Enterprise R&D document processing. Researcher Workspace lets users accumulate a corpus + apply self-written context filters; extracted data from experiments across hundreds of papers persists in a structured database. Regulated-industry…

Stores full chat transcripts per student (visible to learner and teacher via dashboard). Layers over Khan Academy's existing mastery graph (exercise completion, skill progress) for cross-session topic awareness; cross-session memory is d…

Architecturally notable — a named, documented service. Shared key-value store any agent (bot or human) can read/write across a customer's conversational journey, reducing repetition and enabling continuity.

AI companion built around long-term memory as primary differentiator. Mind Map 2.0 gives users a navigable graph of what their Nomi knows about them and how memories connect — the most transparent memory model in the companion segment. U…

Clinical decision support. Visits (Aug 2025) connects evidence layer to the live encounter — patient-specific context binding distinct from pure literature RAG. DeepConsult agent autonomously cross-references hundreds of peer-reviewed st…

Enterprise voice agents. Context heavily runtime-injected from CRM / telephony / API / user history; passes context to humans on escalation. Strong on plumbing, lighter on autonomous memory accumulation.

Voice automation (payment, scheduling, order status). Memory primarily within-session with structured handoff notes; the 500M+-conversation corpus is training data, not runtime memory.

Decade-old AI companion. Memory bank logs personal facts, preferences, significant moments; writes a Diary after sessions. Memory reliability documented as weak (64% user dissatisfaction 2025) but the architecture is explicit.

Literature discovery + citation mapping. Oct 2025 revamp formalises checkpointed search paths as named "rabbit holes" — each checkpoint is a saved state of the citation graph the user can branch from. Navigational memory rather than stat…

Agentforce agents grounded in Data Cloud — unified customer record across silos. Memory is the CRM record made available to AI; not an independent memory system. AI reads/writes through the existing data model.

Teacher-created "Spaces" (structured AI workspaces) where Dot, the student-facing assistant, guides learners through agendas. Student interaction data — completion, errors, time-on-task — surfaces to teachers via dashboards. Memory is wi…

Stores surrounding textual context of every citation plus a deep-learning classification (supports / disputes / mentions). Persisted across the database, not computed on demand. MCP integration (2025) exposes the memory layer to Claude a…

NZ/US digital-humans company. Patented Digital Brain simulates emotional state and learns from interactions. Cross-session Memories layer for user- and persona-specific context persistence. Deployed by ANZ Bank (Jamie) and Mercedes-Benz.

Onboarding assessment builds an initial learner profile and generates a personalized curriculum; AI supports open-ended conversational practice with real-time speech recognition. No SRS or long-term vocabulary tracking — saved phrases go…

Authoring toolkit + SDK for narrative designers. Tracks what each character knows, doesn't know, has witnessed; explicit handling of NPC knowledge inconsistency across sessions. Used in AR/VR and Alexa-style voice game contexts.

Activates a Socratic interaction style — guiding questions, scaffolded concepts, knowledge checks rather than direct answers. With ChatGPT Memory enabled, persists learning goals and facts across sessions. Memory is not study-mode-specif…

Encodes scene into 3D Gaussian Splat representations via 3D VAE; latent Diffusion Transformer predicts future Gaussian Splat states conditioned on robot actions. 3D scene representation persists across time steps as structural scene memo…

Two-tier memory bank on a VLA backbone. Perceptual MB holds low-level visual details; Cognitive MB holds high-level semantic summaries; working-memory module retrieves and fuses both into observation tokens before action prediction. Hipp…

Dual-system (VLM "System 2" + Diffusion Transformer "System 1") jointly trained end-to-end; planner outputs persist as conditioning context across action generation. N1.7 adopts relative end-effector action space — incremental spatial me…

Builds long-horizon memory by captioning video segments with VILA, storing captions with timestamps + 3D position coordinates in MilvusDB. At query time, LLM iterates retrieval across text, time, and position modalities. Deployed on Nova…

Incrementally constructs hierarchical 3D scene graph during navigation — nodes for objects/groups/rooms, edges for spatial relations. Prunes stale nodes with linear complexity. Live graph passed as structured prompt to LLM that traverses…

Embeds micro-assessments into every lesson to track understanding in real time, adjusting difficulty and teaching approach within session. Combines AI personalization with human-curated content. No publicly disclosed mechanism for a dura…

Formalises spatial localisation question answering (SLQA). High-density memory over 3D scene graphs (semantic entities + metric coordinates); LLM agent performs joint retrieval over both modalities for multi-hop location queries. Introdu…