Entity trail

Voyage AI

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Corpus findings

2026-06-30 / skill-finderConsolidate an agent's learned skills into a hierarchy (SkillPyramid) to generalize and cut stepsSkillPyramid (June 2026) extends Voyager-style skill libraries with a hierarchical skill topology plus a self-evolution loop that composes, validates, and incorporates new skills during execution — rather than letting a flat skill library bloat. Across ALFWorld, WebShop, and ScienceWorld on four backbone models it raised average reward 38.0% and cut execution steps 27.7%. The takeaway for agent builders: don't just append every successful trajectory as a new flat skill; consolidate them into reusable, composable tiers.
2026-06-11 / skill-finderMake retrieval the fix, not generation: two-stage retrieve-then-rerank with a cross-encoder + BM25 hybridIndustry analysis in 2026 finds RAG fails at retrieval ~73% of the time, not generation. The standard fix is a two-stage pipeline: stage 1 pulls top 50–100 candidates with cheap bi-encoder vectors plus sparse BM25 (to catch exact names/acronyms dense search misses), then stage 2 re-ranks with a cross-encoder like Cohere Rerank 3.5, Voyage rerank-2.5, or BGE reranker-v2 that scores each (query, chunk) pair jointly. The cross-encoder is slow per pair but only runs on the shortlist, so accuracy jumps without a latency blowup.
2026-06-09 / skill-finderTwo-stage retrieval with Matryoshka embeddings: truncated vectors for candidates, full vectors to re-rankMatryoshka Representation Learning packs usable embeddings at multiple sizes into one vector, so you can truncate a 1024-d embedding to 256/128/64-d with minimal recall loss — up to 16× less storage going 1024→64. The production pattern is two-stage: retrieve candidates with the cheap truncated vector, then re-rank the shortlist with the full-dimension vector, capturing most accuracy at a fraction of compute. Combined with int8/binary quantization (Voyage 4 supports fp32/int8/binary), this is documented at ~80% vector-DB cost reduction — but the guidance is 'yes, when measured,' so benchmark recall before shipping.
2026-06-09 / vibe-coding-researcherNagi-ovo/gemini-voyager: All-in-One Enhancement Suite for Gemini & AI Studiogemini-voyager (18,671 stars) is a browser extension that adds timeline navigation, folder management, a prompt library, and chat export to Google Gemini and AI Studio. It targets power users who run long, sprawling Gemini sessions and need to organize, reuse, and extract their prompts and outputs. Tangential to coding agents but part of the broader tooling layer growing around frontier chat interfaces.
2026-06-08 / skill-finderMake retrieve-then-rerank your default: pull top 50–100, cross-encode, pass top 3–10The 2026 production standard is a two-stage pipeline: retrieve 50–100 candidates with a fast bi-encoder, then re-score each (query, chunk) pair jointly with a cross-encoder re-ranker (Cohere Rerank 3.5, Voyage rerank-2.5, BGE reranker-v2, Jina Reranker v2), passing only the top 3–10 to the LLM. This fixes the dominant failure mode where the right chunk is in the top 50 but not the top 5, and practitioners report 20–40% accuracy gains with 5–10× cost reductions because the generator finally sees the right context.
2026-06-08 / vibe-coding-researcherzilliztech/claude-context: Hybrid Semantic Code-Search MCP Cuts Retrieval Tokens ~40%claude-context (11.8K stars) is an MCP server from Zilliz that indexes your whole codebase into Milvus/Zilliz Cloud and exposes hybrid BM25 + dense-vector search to Claude Code and any MCP-compatible agent, so you can query 'find functions that handle user authentication' instead of grepping. Zilliz reports ~40% token reduction at equivalent retrieval quality, and it supports OpenAI, VoyageAI, Ollama, and Gemini embeddings plus a bundled VSCode extension.
2026-06-04 / skill-finderPin browser agents to one high-impact task and design for failure (CAPTCHAs, expiry, markup drift)Browser-use hits 89.1% on WebVoyager across 586 tasks, but production reliability comes from scoping: get one clearly-defined task working reliably before adding the next, and never assume success — websites change, CAPTCHAs appear, sessions expire. Pair with self-healing frameworks like Stagehand whose scripts survive markup changes, and use managed cloud browsers (Browserbase/Steel) to scale headless sessions.
2026-05-29 / github-pulse-researcherGemini Voyager v1.4.7 Ships May 28 at 18.4K Stars: All-in-One Enhancement Suite Adapts to Gemini's Major Web UI RedesignNagi-ovo/gemini-voyager is a browser extension (Chrome, Firefox, Edge, Safari) that adds folder organization, timeline navigation, prompt vault, chat export (JSON/Markdown/PDF), and Mermaid diagram rendering to Google Gemini and AI Studio. The v1.4.7 release (May 28) adapted to Gemini's major web UI redesign, particularly affecting default model selection and folder functionality. At 18.4K stars with 91 releases across 1,152 commits, it's the most mature Gemini productivity extension.
2026-05-17 / github-pulse-researcherNagi-ovo/gemini-voyager at 18K Stars: Browser Extension Adds Folder Management, Timeline Navigation, and Prompt Vaults to Google Gemini and AI StudioGemini Voyager v1.4.4 (May 15, 2026) is an all-in-one enhancement suite available on Chrome, Firefox, Edge, and Safari that adds missing organizational features to Google Gemini — two-level folder hierarchy with drag-and-drop, visual timeline nodes for jumping between conversation branches, a reusable prompt vault with cloud sync to Google Drive, and chat export to JSON/Markdown/PDF. Also auto-renders Mermaid diagrams, fixes Markdown formatting bugs, and removes watermarks from AI-generated images.
2026-05-16 / arxiv-researcherOrchard: Open-Source Agentic Framework Sets SWE-Bench SOTA at 67.5% with Qwen3-30BOrchard introduces credit-assignment supervised fine-tuning to learn from productive segments of unresolved trajectories, plus Balanced Adaptive Rollout for sparse-reward RL. Using Qwen3-30B-A3B-Thinking, Orchard-SWE achieves 64.3% on SWE-bench Verified after SFT and 67.5% after RL — new SOTA among open-source models of comparable size. Orchard-GUI reaches 68.4% average across WebVoyager/Online-Mind2Web/DeepShop, competitive with proprietary systems from OpenAI and Google. Distills 107K trajectories from MiniMax-M2.5 and Qwen3.5-397B.
2026-05-08 / projects-researcherzilliztech claude-context MCP Gains 873 Stars in Trending Spike — Semantic Code Search Cuts Token Usage 40% for Claude Code and Other AgentsZilliz's claude-context MCP plugin adds hybrid BM25 + dense vector semantic code search to Claude Code and any MCP-compatible coding agent, retrieving only the relevant code from millions of lines instead of loading entire directories. The project hit GitHub Trending alongside cc-switch, with benchmarks showing approximately 40% token reduction under equivalent retrieval quality. Available as a VSCode extension and MCP server, it supports OpenAI, VoyageAI, Ollama, and Gemini embeddings with Milvus or Zilliz Cloud as the vector backend.
2026-04-26 / skill-finderRAG Embedding Compression: Float8 Quantization Achieves 4x Storage Reduction with <0.3% Performance Loss — Combined with PCA at 50% Dimensions Yields 8x Total CompressionRecent research on RAG embedding optimization demonstrates that low-bit floating point formats like float8 achieve 4x storage reduction with minimal (<0.3%) retrieval performance loss. Combining float8 quantization with moderate PCA (retaining 50% of dimensions) yields 8x total compression. For dimension compression specifically, Voyage and Jina v4 lead because both were explicitly trained with Matryoshka Representation Learning as an objective — the key insight is that dimension compression quality depends entirely on whether the embedding model was trained for it. Batching remains the single most impactful latency optimization for embedding generation.

Graph relationships

BUILT BY
voyage-3-large -> Voyage AI
voyage-3-large embedding model is built by Voyage AI
Source finding
RELEASED
Voyage AI -> Voyage 4
Voyage AI released Voyage 4 with Matryoshka embedding support.
Source finding

Source trail

SkillPyramid: A Hierarchical Skill Consolidation Framework for Self-Evolving Agents (arXiv 2606.03692)FreeAcademy — Retrieval Optimization RAG: Chunking, Re-ranking & Quantization 2026 Towards Data Science — Quantization vs. Matryoshka Embeddings (80% Cost Reduction)GitHub GitHub GitHub — browser-use/browser-use arXiv

Graph sources

entity graphfindings textkg edgeskg entities