# US 12,602,549 B1 · Persistent cognitive machine with curated long term memory > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 12,602,549 B1 - Document ID: US12602549B1 - Status: Granted - Kind: B1 - Application number: 19/321,173 - Filed: 2025-09-06 - Granted: 2026-04-14 - Earliest claimed: 2025-02-12 - Assignees: Atombeam Technologies Inc. - Applicants: AtomBeam Technologies Inc. - Inventors: Brian Galvin - Web page: https://patentgrep.com/patent/US12602549B1 ## CPC - G06F 40/30: Semantic analysis - G06F 16/3325: {Reformulation based on results of preceding query} - G06F 16/3329: Natural language query formulation - G06F 16/90332: {Natural language query formulation or dialogue systems} - G06N 3/006: based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO] - G06N 3/042: Knowledge-based neural networks; Logical representations of neural networks - G06N 3/044: Recurrent networks, e.g. Hopfield networks - G06N 3/045: Combinations of networks - G06N 3/0455: Auto-encoder networks; Encoder-decoder networks - G06N 3/0464: Convolutional networks [CNN, ConvNet] - G06N 3/047: Probabilistic or stochastic networks - G06N 3/049: Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs - G06N 3/063: using electronic means - G06N 3/08: Learning methods - G06N 3/084: Backpropagation, e.g. using gradient descent - G06N 3/088: Non-supervised learning, e.g. competitive learning - G06N 3/09: Supervised learning - G06N 5/01: Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound - G06N 5/02: Knowledge representation; Symbolic representation - G06N 5/022: Knowledge engineering; Knowledge acquisition - G06N 5/04: Inference or reasoning models - G06N 5/041: {Abduction} - G06N 7/01: Probabilistic graphical models, e.g. probabilistic networks - G06N 20/00: Machine learning - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) - G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS ## Abstract A system and method for implementing persistent cognitive computation through geometric representation of thought in a dynamic latent manifold. The system encodes inputs into a curved space characterized by time-evolving metric tensors, compression pressure fields derived from Ricci curvature, and goal potential fields that shape attention flow. Cognition occurs through geodesic traversal of this manifold, with attention following paths that minimize cognitive action while balancing semantic density and goal relevance. A Cognitive Dynamics Engine maintains manifold geometry, computing optimal trajectories and managing thought bundle operations including consolidation, expansion, and higher-order abstraction. During idle periods, autonomous dreaming processes reorganize the manifold through perturbation, recombination, and topological surgery. This architecture enables persistent memory through geometric encoding, where frequently accessed concepts develop high-curvature regions and cognitive shortcuts emerge from usage patterns, transforming artificial intelligence from stateless computation to structured motion through shaped memory space. ## Claims ### Claim 1 A computer system comprising a hardware memory, wherein the computer system is configured to execute software instructions stored on nontransitory machine-readable storage media that: maintain a latent manifold as a geometric substrate for cognitive operations, wherein the latent manifold evolves through use; encode inputs into geometric structures within the latent manifold, wherein semantic relationships are represented through geometric properties including distance and curvature; compute paths through the latent manifold for cognitive processing, wherein the paths are influenced by the geometric structure of the manifold; store persistent representations as geometric regions within the latent manifold, wherein frequently accessed representations develop characteristic geometric properties that facilitate future access; modify the geometric structure of the latent manifold based on cognitive operations, wherein successful reasoning patterns create persistent modifications to the manifold geometry; and generate outputs by traversing the latent manifold and decoding geometric information into user-interpretable responses. ### Claim 2 (depends on 1) The computer system of claim 1, wherein the software instructions further: compute compression pressure fields derived from local curvature of the latent manifold, wherein regions of high semantic density exhibit higher compression pressure that influences path computation. ### Claim 3 (depends on 1) The computer system of claim 1, wherein the software instructions further: organize persistent representations into thought bundles comprising coherent submanifolds of semantically related concepts, wherein the thought bundles support operations including consolidation, expansion, and recombination. ### Claim 4 (depends on 1) The computer system of claim 1, wherein the software instructions further: execute autonomous reorganization of the latent manifold during idle periods, including perturbation of existing structures, synthesis of new connections between disparate regions, and removal of unused or redundant structures. ### Claim 5 (depends on 1) The computer system of claim 1, wherein the software instructions further: implement a distributed thought cache that stores frequently accessed geometric structures, wherein cache hits enable direct response generation without full path computation through the latent manifold. ### Claim 6 (depends on 1) The computer system of claim 1, wherein the software instructions further: track activation energy for each persistent representation, wherein representations with low activation energy undergo thermodynamic decay and eventual removal from the latent manifold. ### Claim 7 (depends on 1) The computer system of claim 1, wherein the software instructions further: maintain bidirectional attention fields that support both forward exploration toward goals and reverse traversal along previously computed paths, enabling backtracking and path revision. ### Claim 8 (depends on 1) The computer system of claim 1, wherein the software instructions further: establish goal potential fields that create attractive forces within the latent manifold, guiding path computation toward semantically relevant regions for achieving specific objectives. ### Claim 9 (depends on 1) The computer system of claim 1, wherein the software instructions further: implement hierarchical organization with multiple nested latent manifolds operating at different levels of abstraction, wherein paths can traverse between abstraction levels through geometric bridges. ### Claim 10 A method for a persistent cognitive computation through geometric representation of thought in a dynamic latent manifold, comprising the steps of: maintaining a latent manifold as a geometric substrate for cognitive operations, wherein the latent manifold evolves through use; encoding inputs into geometric structures within the latent manifold, wherein semantic relationships are represented through geometric properties including distance and curvature; computing paths through the latent manifold for cognitive processing, wherein the paths are influenced by the geometric structure of the manifold; storing persistent representations as geometric regions within the latent manifold, wherein frequently accessed representations develop characteristic geometric properties that facilitate future access; modifying the geometric structure of the latent manifold based on cognitive operations, wherein successful reasoning patterns create persistent modifications to the manifold geometry; and generating outputs by traversing the latent manifold and decoding geometric information into user-interpretable responses. ### Claim 11 (depends on 10) The method of claim 10, further comprising the step: computing compression pressure fields derived from local curvature of the latent manifold, wherein regions of high semantic density exhibit higher compression pressure that influences path computation. ### Claim 12 (depends on 10) The method of claim 10, further comprising the step: organizing persistent representations into thought bundles comprising coherent submanifolds of semantically related concepts, wherein the thought bundles support operations including consolidation, expansion, and recombination. ### Claim 13 The method of 10, further comprising the step: executing autonomous reorganization of the latent manifold during idle periods, including perturbation of existing structures, synthesis of new connections between disparate regions, and removal of unused or redundant structures. ### Claim 14 (depends on 10) The method of claim 10, further comprising the step: implementing a distributed thought cache that stores frequently accessed geometric structures, wherein cache hits enable direct response generation without full path computation through the latent manifold. ### Claim 15 (depends on 10) The method of claim 10, further comprising the step: tracking activation energy for each persistent representation, wherein representations with low activation energy undergo thermodynamic decay and eventual removal from the latent manifold. ### Claim 16 (depends on 10) The method of claim 10, further comprising the step: maintaining bidirectional attention fields that support both forward exploration toward goals and reverse traversal along previously computed paths, enabling backtracking and path revision. ### Claim 17 (depends on 10) The method of claim 10, further comprising the step: establishing goal potential fields that create attractive forces within the latent manifold, guiding path computation toward semantically relevant regions for achieving specific objectives. ### Claim 18 (depends on 10) The method of claim 10, further comprising the step: implementing hierarchical organization with multiple nested latent manifolds operating at different levels of abstraction, wherein paths can traverse between abstraction levels through geometric bridges.