Base Layer is an open-source reference pipeline that produces a Behavioral Specification of how a specific person interprets information, decides, and communicates. The Specification is one implementation of an interpretive layer above memory: the framework an AI agent reads facts through to act in alignment with the individual rather than the population average.
base-layer.ai · Live examples · Research · Benchmark dataset
Current AI memory systems (Mem0, Letta, Supermemory, Zep) optimize for recall and score 70 to 93 percent on standard benchmarks (LOCOMO, LongMemEval). Recall is approaching saturation on those benchmarks.
What's left unmeasured is interpretation: how a specific person turns facts and experiences into judgments, decisions, and reactions. Facts are inert until something reads them, and the lens should be the individual's. The Behavioral Specification is one implementation of that missing interpretive layer, and the artifact the Beyond Recall paper tests. The category claim, that interpretation is a measurable axis distinct from recall, is the contribution. The specific artifact is the demonstration. Other implementations of an interpretive layer are welcome and expected.
A structured document encoding a person's behavioral patterns across three interpretive layers plus a composed unified brief. About 7,000 tokens, roughly 5,000 words, the length of a short magazine article. (Paper §3.7.)
ANCHORS Decision foundations. The axioms someone reasons from.
CORE Operational constraints. Communication patterns and context modes.
PREDICTIONS Behavioral triggers with detection cues and directives.
Generated from raw text through a five-step pipeline:
IMPORT Multi-source ingest (ChatGPT, Claude, journals, text) -> SQLite
EXTRACT Haiku, 46 constrained behavioral predicates -> structured facts
EMBED MiniLM-L6-v2 local embeddings -> ChromaDB vectors
AUTHOR Sonnet, three-layer authoring with domain-agnostic guard -> anchors / core / predictions
COMPOSE Opus, compresses three layers into one specification -> ~7K token document
baselayer run <file> executes the full pipeline with a cost-estimate gate. See docs/core/ARCHITECTURE.md for the canonical description of each stage.
First paragraph of a real Specification, generated from approximately 1,900 conversations:
He operates from an uncompromising need for logical coherence that manifests as immediate challenge to any inconsistency, in systems, arguments, or his own positions. When he encounters a gap between stated beliefs and actual behavior, he treats it as personal failure requiring accountability rather than understanding, taking extreme ownership of every outcome while maintaining clear causal links between actions and results. This isn't philosophical posturing but lived practice: in trading, he waits for multiple confirming signals before entries, implements overlapping safety mechanisms through fixed dollar loss limits and systematic stop losses, yet struggles with the gap between knowing these rules and executing them consistently during early morning sessions when his energy is highest but discipline most vulnerable.
Text alone. No questionnaires, no profiles, no manual input. Every claim cites the facts it was authored from, and every fact cites the source passage it was extracted from. More examples.
The Specification gives a model person-specific grounding where it would otherwise refuse or guess from population averages, and it helps most where the model knows the person least. It composes above other context rather than replacing it: memory systems and raw text supply facts, the Specification supplies the lens those facts are read through, and that lens travels with the person across any model or provider. It is small enough to serve on every turn yet carries most of the predictive signal of a raw corpus many times its size.
It also acts as a leveler. On a subject a model already knows well from pretraining it adds little, but on the long tail of people whose reasoning sits in no training corpus, it brings responses to roughly the same grounded, person-specific quality. The full evaluation across 14 public-domain subjects, with numbers and methodology, is in the Beyond Recall paper and at base-layer.ai/research.
- Not memory on the retrieval axis. Memory systems retrieve facts; the Specification supplies interpretation. Different layers.
- Not a competitor to Mem0, Letta, Supermemory, or Zep on recall benchmarks. Those systems perform within a few points of each other; the Specification operates at a different layer.
- Not magic on subjects the model already knows well. High-baseline subjects show near-zero or mildly negative lift.
- Not a final implementation. The Behavioral Specification is one implementation of the interpretive layer; others are welcome and expected. (Paper §1.4.)
Requirements: Python 3.10+, Anthropic API key.
pip install git+https://github.com/agulaya24/BaseLayer.git
export ANTHROPIC_API_KEY=sk-ant-...
baselayer run chatgpt-export.zipBase Layer is not yet on PyPI; the
baselayername is held by an unrelated project. Install from source via the URL above, or clone the repo andpip install -e ..
Runs the full pipeline with a cost gate. Roughly 30 minutes and $0.50 to $2.00 for ~1,000 conversations.
For step-by-step control:
baselayer init
baselayer import chatgpt-export.zip # or claude-export.json, ~/journals/, notes.md
baselayer estimate # preview cost before spending
baselayer extract && baselayer embed
baselayer author && baselayer composeOther input types. Books, essays, letters, patents: baselayer extract --document-mode. No conversation history? baselayer journal runs guided prompts that bootstrap a starter Specification.
Cloud dependency. Extraction, authoring, and composition send text to the Anthropic API (zero-retention by default). Extraction can run fully local via Ollama; authoring and composition currently require the Claude API.
Two ways to put a Specification in front of a model: register the MCP server, or paste the unified brief into a system prompt.
MCP server (Claude Desktop, Claude Code, Cursor):
claude mcp add --transport stdio base-layer -- baselayer-mcpReads from the same SQLite + ChromaDB store the pipeline builds, with no re-indexing. It loads an always-on resource (memory://specification, approximately 6 to 8K tokens: CORE, ANCHORS, PREDICTIONS, plus a tools manifest) and on-demand tools the model calls when it needs grounded retrieval or provenance:
| Tool | Purpose |
|---|---|
get_brief(reason) |
Unified narrative portrait of the user (~3,000 tokens). |
recall_memories(query) |
Semantic retrieval over facts and episodes (ChromaDB, MiniLM-L6-v2). |
search_facts(query, limit) |
FTS5 keyword search across active facts. |
trace_claim(claim_id) |
Provenance from a claim (e.g. A1, P3) back to source facts. |
verify_claims(claim_id, layer) |
Verification checks against the fact database. |
get_stats() / get_call_log() / get_help(topic) |
Database summary, session calls, agent reference. |
Stdio, local, no network. Per-session traces land in ~/.baselayer/sessions/<pid>/log.jsonl (baselayer log list / show / tail / stats).
Paste directly. Paste the full Specification (three layers plus unified brief, approximately 7,000 tokens) into Claude custom instructions, ChatGPT project files, or any system prompt. Keeps the structural Specification, loses the on-demand fact retrieval.
Every claim in a generated Specification cites the facts used to author it, and every fact cites the source passage it was extracted from. baselayer verify runs four checks against that citation graph: vector proximity (topic consistency), recurrence gating (no claim stands on a one-off mention), cross-domain span (no single-domain overfit), and NLI entailment (a local DeBERTa model scores supportability). This is a strong data-quality audit, not a causal-traceability guarantee. You can inspect every claim's evidence chain and flag low-recurrence or single-domain citations.
A methods paper plus an open-source reference pipeline, not a product launch. The category claim is the contribution; the specific artifact is the demonstration.
- Tested on 14 historical subjects with public-domain autobiographies (4 continents, ~2 millennia of written experience). 5-judge primary panel, 7-judge sensitivity, pre-registered analysis plan. Full results: base-layer.ai/research.
- Direction reproduces across response models and battery-generation models; absolute magnitudes are panel-dependent.
- Snapshot, not longitudinal. A point-in-time cross-section.
- Text-only. Tone of voice, body language, and physical habits are invisible to the extractor.
- Cloud dependency for authoring and composition (extraction can run local via Ollama).
- Pre-1.0. 451 tests; expect rough edges.
- Faithfulness is the central open question. A Specification that serves cheaply and scores well on a held-out battery does not entail it structurally matches a person's reasoning. (Paper §5.6.)
The database (SQLite), vectors (ChromaDB), extracted facts, and Specification all live on the user's machine. No cloud sync, no accounts, no telemetry. A representation of how a specific person interprets information is operationally consequential, so Base Layer ships local inspection and modification tools. A representation that is opaque to the person it represents is built for someone else; that is not the design here.
- Benchmark dataset. The Beyond Recall specifications, batteries, facts, corpora, and results are on Hugging Face:
agulaya24/beyond-recall. - Live example Specifications (no auth):
GET https://base-layer.ai/api/identity/{franklin,buffett,douglass}return structured JSON (anchors, core, predictions, unified Specification, stats). - For agents and LLMs:
llms.txt,llms-full.txt, Agent card, MCP server card, OpenAPI spec.
| Doc | Contents |
|---|---|
ARCHITECTURE.md |
Pipeline design, canonical 5-step description |
PROJECT_OVERVIEW.md |
Internal architecture, components, composition |
DECISIONS.md |
Design decisions with rationale |
DESIGN_PRINCIPLES.md |
Foundational principles |
ROADMAP.md |
Near-term, mid-term, and research-horizon work |
docs/eval/ |
Evaluation frameworks and study results |
The prompts are in the code. Nothing is hidden.
The repository version corresponding to the Beyond Recall paper is tagged v0.2.0, and a frozen copy is vendored into memory-study-repo under ./baselayer/. Old surfaces (the /api/identity/{subject} URLs, the memory://identity MCP URI, the --identity-only CLI flag) continue to serve indefinitely as aliases; new names are added alongside, never as replacements.
pip install git+https://github.com/agulaya24/BaseLayer.git@v0.2.0Contributions welcome, especially around evaluation, source-type adapters, alternative interpretive-layer implementations, and local model support. See CONTRIBUTING.md.
@software{baselayer2026,
title = {Base Layer: An Open-Source Reference Pipeline for the Interpretive Layer Above Memory},
author = {Gulaya, Aarik},
year = {2026},
url = {https://github.com/agulaya24/BaseLayer},
license = {Apache-2.0}
}Apache 2.0. See LICENSE. The accompanying Beyond Recall paper is CC-BY 4.0.