Competitive Positioning

Zaxy's product thesis is "git for agent memory": an event-sourced, replayable, auditable memory fabric that projects durable context into graph, lexical, verbatim, and active working-set views.

What Stands Out

Eventloom append-only JSONL is the source of truth, with hash-chain integrity and deterministic replay.
Bi-temporal graph projection preserves what was true then versus now.
Retrieval uses graph, lexical, vector, verbatim source recall, and active working-set projection rather than only chunk similarity.
Automatic capture writes typed lifecycle observations for transcript turns, tool calls, command results, file edits, compaction, subagents, and session boundaries.
Context assembly returns prompt-ready memory with citations, source lanes, policy metadata, warnings, and feedback hooks.

MemPalace Target

MemPalace and AgentGraph-style graph memory are the current public validation points for Zaxy's thesis. The market is moving toward persistent, structured memory rather than flat context files. Zaxy should compete on trust, provenance, and replayability rather than trying to match every UX surface first. The benchmark lane should stay architecture-driven:

Temporal correctness: recover old and current facts without overwriting history.
Source recall: answer with verbatim Eventloom citations and transcript source anchors.
Relational recall: follow graph relationships across goals, tasks, decisions, files, symbols, and test coverage.
Context collapse resistance: preserve identity through compaction and active working-set projection.
Auditability: replay how a memory was written, projected, retrieved, and reinforced.

Public Benchmark Posture

The current Zaxy public benchmark hub is benchmarks.md. Zaxy's same-harness evidence now leads with the full 500-question current74 checkout archive: mean score 0.940, Answer@5 0.906, citation coverage 1.000, R@1/R@5/R@10 of 0.906/1.000/1.000, p95 687.67 ms, and p99 969.10 ms on workload SHA-256 90fb2307195d7e16b963a2b8a30f03b375bd42a45d41aeaa55423029dd84e3fc. The 100-question BM25 comparison remains a useful tradeoff report: Zaxy mean score 0.970, Answer@5 0.950, citation coverage 1.000, and R@1/R@5/R@10 1.000, with BM25 in the same report at mean score 0.540, Answer@5 0.500, and R@5 0.840. The older BM25-included full 500-question hash run remains historical legacy limit=10 floor evidence, not the current headline. Projection-backend work should use the current same-harness limit=5 backend-evaluation floor instead of mixing those older limit=10 thresholds into backend comparisons.

Competitor numbers belong in an external-disclosure table, not a universal leaderboard. MemPalace publicly reports 96.6% raw LongMemEval R@5 and 98.4% held-out hybrid R@5, plus optional LLM-reranked full-set runs reported at 99%+ R@5. Agent Memory publicly reports 95.2% R@5 on LongMemEval-S. Mem0 publicly reports LongMemEval accuracy in the low-to-mid 90s and lower-token memory retrieval, plus LoCoMo accuracy gains. Zep/Graphiti remains an important graph-memory product reference point, but its public numbers should still be kept separate unless a same-harness adapter is pinned. GBrain is the benchmark-visible Harvey LAB comparison: Zaxy's audited Harvey run beats the article-best task rows by +0.081 mean criterion pass rate and wins 9/10 tasks, but that is article-relative evidence rather than a Zaxy-run GBrain adapter. These claims are important market context, but they are not same-harness Zaxy results.

Same-Harness Adapter Feasibility

MemPalace is the strongest adapter candidate because its public repo documents a local benchmarks/longmemeval_bench.py path and committed per-question result files. A Zaxy adapter should wrap that command, pin the mode and top-k settings, and import per-question retrieval hits into Zaxy's report schema.

Mem0 is a benchmark harness candidate rather than a drop-in retrieval adapter: the public mem0ai/memory-benchmarks project can run LongMemEval, but the OSS path requires Docker, Qdrant, model configuration, and LLM answer/judge choices. The first Zaxy integration should document those inputs and separate retrieval-only comparisons from judge-scored answer accuracy.

Agent Memory remains external disclosure only for now. Its product page reports LongMemEval-S R@5 and a BM25/vector/graph retrieval stack, but the public page does not provide a stable same-harness CLI/API contract for Zaxy to call. Keep the number in the disclosure table until a reproducible command and result export are available.

Zep/Graphiti should be treated as a larger graph-memory comparison target, not as a thesis-adjacent ally to attack. A same-harness adapter would need to pin the graph construction path, retrieval limit, source-citation mapping, and latency/token accounting.

GBrain belongs in the Harvey LAB lane. The current publishable claim is that Zaxy beats the article-best GBrain task rows in the audited Harvey artifact comparison; a stronger claim requires a pinned GBrain runner.

Near-Term Roadmap

Treat the current74 full 500-question LongMemEval-compatible report as the public headline and no-regression floor: mean score 0.940, Answer@5 0.906, citation coverage 1.000, and R@5 1.000. Work down synthesis misses without reducing citation coverage or R@5.
Keep same-harness BM25 baselines in public benchmark reports so Zaxy claims show both retrieval quality and latency/token tradeoffs against a strong lexical baseline.
Build or document feasible same-harness competitor adapters for MemPalace, Mem0, Agent Memory, and Zep/Graphiti. Keep GBrain in the Harvey LAB article-relative lane until a pinned runner exists. If a same-harness adapter is not practical, document the blocker and keep the competitor number in the external-disclosure table.
Add Skill Memory as a first-class procedural layer: skill lifecycle events, skill-version graph projection, checkout skill routing, outcome tracking, and eval-gated promotion/rollback.
Evaluate pgGraph as an experimental Postgres-backed projection backend behind a backend-neutral contract, with embedded Kuzu remaining the default until a candidate backend proves same-harness benchmark and operations parity or better.
Expand clean-repo onboarding and model-facing UAT across fresh Codex and Claude Code workspaces: zaxy init, Memory Bootstrap, deterministic capture, Memory Checkout, retrieval feedback, and doctor status.
Keep public benchmark scripts reproducible from a clean checkout with cached datasets, archived reports, guardrail commands, and no hidden manual steps.
Keep LangGraph native-beta, CrewAI native-preview, and OpenAI-compatible model-call adapters dependency-light on the shared native payload contract, zaxy.native.v0.6.
Keep AutoGen template-only until the right runtime hooks are clear.
Add public comparison workloads that evaluate temporal, source-recall, graph-traversal, and context-collapse behavior against MemPalace-style memory.
Keep claims reproducible: every comparison should publish workload hashes, retrieval settings, and exact scoring rules.