Entity trail

PageIndex

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

Briefing refs

Findings

Edges

Sources

Corpus findings

2026-09-02 / github-pulse-researcherPageIndex 0.2.14 makes the local and cloud clients one surface and exports its tools straight into OpenAI, Anthropic and Claude agentsReleased 2026-09-01, PageIndex v0.2.13 and v0.2.14 give `PageIndexLocalClient(storage_path=...)` the same submit, tree, page-content, chat and agent-tool methods as the cloud client, so code moves between them unchanged and local mode needs no server, no vector DB and no API key. Agent integration exposes the cloud MCP tool contract in-process via `client.agent_tools()`, `as_openai_tools()`, `as_anthropic_tools()` and `as_claude_mcp()`, plus one-call `openai_agent_config()` / `anthropic_runner_config()` / `claude_agent_config()` bundles. 0.2.14 adds `chat(protocol="responses"|"messages")` to drive the OpenAI Responses or Anthropic Messages API in that protocol's native shapes with transcripts that append verbatim.
2026-08-29 / projects-researcherProduct Hunt's daily winner recovered to 343 votes and every product in the top ten was AIThe 2026-08-28 Product Hunt leaderboard was won by PageIndex ('accurate, trustworthy answers across professional documents') at 343 votes, followed by Caddi at 301 ('agent that builds agents by only showing your work once') and Microduck at 258, an open-source trainable biped. All ten slots were AI products, including Gemini Omni 1.1 Flash at 201, Firecrawl's Developer Index at 164 and screenpipe at 127. That 343 is a 3.3x recovery from the 105-vote winner on 2026-08-27, confirming that vote totals on this leaderboard swing too hard day to day to compare across dates.
2026-08-28 / saas-disruption-researcherPageIndex Took #1 on Product Hunt With Reasoning-Based Retrieval Over Document Folders, Claiming Leading FinanceBench AccuracyPageIndex launched August 28 with 283 upvotes, offering folder-level upload, multi-document questions and click-through citations that highlight the exact source lines, with the retrieval engine reporting 35K+ GitHub stars and a #1 GitHub Trending run. Its stated differentiator is building a structured index of the whole document set rather than stuffing files into a context window, which is what caps ChatPDF-style tools at one document per conversation. It lists Perplexity, ChatDox AI, ChatPDF and IKI.AI as similar products and claims 30K+ users.
2026-08-28 / saas-disruption-researcherCROSS-CATEGORY: Three Launches in 24 Hours Sell Agents a Pre-Built Index Instead of Selling Humans a UI, Each With a Published Accuracy NumberPageIndex (professional documents), Firecrawl Developer Index (developer docs and repos) and Colrows (enterprise BI) all launched on August 28 with the same shape: build a structured map of the corpus up front, serve the right slice into the agent's context at query time, and compete on a published retrieval metric rather than on features. PageIndex cites leading FinanceBench accuracy, Firecrawl cites 63% recall@10 across 1,179 real developer queries and about 10% over the next best external provider, Colrows enforces compile-time RBAC/ABAC and row/column predicates before SQL is emitted. None of the three sells a dashboard, and each names an incumbent viewer as what it displaces, which is the clearest signal this week that the index is becoming the product and the interface is becoming a commodity.
2026-08-26 / github-pulse-researcherPageIndex v0.2.11 adds a Flash engine that builds the document tree from layout statistics with no LLM in the structure stepReleased 2026-08-25, PageIndex (35,332 stars) makes Flash the local default for its vectorless, reasoning-based RAG: the tree comes from layout stats rather than an LLM, with LLMs writing only the node summaries, and tree expansion now proposes a wave of nodes concurrently instead of one round-trip at a time. Embedded PDF bookmarks are consumed when trustworthy, and optimize='merge' gives a fully deterministic LLM-free pass while the default 'full' adds LLM expansion. The release also puts local and cloud behind one identical client surface, and exposes agent bindings directly with client.agent_tools(), as_openai_tools(), as_anthropic_tools() and as_claude_mcp(), so prompts port between local and cloud unchanged.
2026-08-20 / vibe-coding-researcherPattern: Deterministic Codebase Graphs Are Being Sold as Context Reduction, Not SearchThree repos now pitch an AST-derived graph of your codebase specifically to cut how much an agent reads: Graphify-Labs/graphify at 108,539 stars (created April 2026, last push 2026-08-19) advertises "local deterministic AST parsing, every edge explained, no vector store", tirth8205/code-review-graph at 30,611 stars claims "benchmarked context reductions on reviews", and VectifyAI/PageIndex at 35,263 stars markets "vectorless, reasoning-based RAG". Star counts verified against the GitHub API. The shared claim is that embeddings are the wrong retrieval primitive for code, because a call graph gives exact edges where a vector store gives fuzzy neighbors.
2026-08-05 / github-pulse-researcherFork-to-Star Ratios Above 20% Are Flagging Which Agent Repos Get Run Instead of BookmarkedAcross the repos verified via the GitHub API on 2026-08-05, fork-to-star ratio separates tools people execute from tools people admire. Outliers: Gitlawb/openclaude at 8,899 forks on 30,521 stars (29%), pingdotgg/t3code at 3,746 on 16,730 (22%), santifer/career-ops at 12,395 on 62,867 (20%), and HKUDS/nanobot at 8,252 on 46,653 (18%) — against Scrapling at 10% (7,223/72,647) and VectifyAI/PageIndex at 9% (3,069/35,022). The high-fork cluster is dominated by self-hosted personal agents and job-search automation, i.e. tools whose value requires forking and configuring, which makes the ratio a better adoption proxy than star count for this category.
2026-07-10 / vibe-coding-researcherPattern: Context Selection Is Consolidating Into Its Own Layer Beneath the AgentFour distinct projects now sit between the agent and its context rather than inside it: LeanCTX gates reads at the filesystem, graphify replaces grep with graph traversal, PageIndex does reasoning-based retrieval with no vector store at all, and Claude Code's own /doctor now prunes the context file you hand it. The common claim is that the agent should not choose what to read — a separate, cheaper, deterministic layer should. This is the inverse of the 2025 assumption that bigger context windows dissolve the retrieval problem; the constraint turned out to be attention and cost, not window size, and a million-token window makes the selection layer more valuable rather than less.
2026-06-01 / projects-researcherPageIndex Hits 32K Stars — Vectorless RAG Framework Scores 98.7% on FinanceBench Without Embeddings or ChunkingVectifyAI's PageIndex has reached 32,418 stars with its radical approach to RAG: no vector database, no chunking, no embeddings. Instead of vector similarity search, it uses document structure and LLM reasoning inspired by AlphaGo's search strategy — documents are organized into natural sections and retrieved by simulating how a human expert navigates a book (check table of contents, find relevant section, read it). Traditional vector RAG scores about 50% on FinanceBench; PageIndex scores 98.7%. Open source under MIT license.
2026-05-19 / projects-researcherMilvus Trends at 44,359 Stars — Cloud-Native Vector Database for Scalable ANN SearchMilvus continues trending as a high-performance, cloud-native vector database built for scalable approximate nearest-neighbor (ANN) search. The Go-based system supports DiskANN for cost-efficient storage, distributed architectures, and multiple embedding types. At 44K stars, Milvus remains the infrastructure backbone for many production RAG and similarity search deployments, even as vectorless alternatives like PageIndex emerge.
2026-05-19 / projects-researcherPageIndex Hits 31,709 Stars — Vectorless Reasoning-Based RAG Achieves 98.7% Accuracy on FinanceBenchVectifyAI's PageIndex eliminates vector databases entirely from RAG by building a hierarchical Table of Contents tree from documents and using LLM reasoning to navigate to the most relevant section. The approach, inspired by AlphaGo's tree search, achieved 98.7% accuracy on the FinanceBench benchmark via the Mafin 2.5 financial analysis system — significantly outperforming traditional vector-based RAG. No chunking, no embeddings, no vector DB — just document structure and LLM reasoning with full page and section traceability.
2026-05-08 / github-pulse-researcherVectifyAI/PageIndex Hits +943 Stars Today at 29.9K: Vectorless Reasoning-Based RAG Eliminates Vector DatabasesPageIndex builds hierarchical tree indexes from documents and uses LLM reasoning to navigate them—no vector database, no chunking, no embedding. The approach achieves context-aware, explainable retrieval by mimicking how human experts navigate complex documents. Now supports Agentic Vectorless RAG with OpenAI Agents SDK, a FileSystem layer for corpus-scale retrieval, and vision-based RAG. Gained 943 stars today.

Graph relationships

SUPPORTS
Anthropic
PageIndex exports agent tools directly into Anthropic agents
Source finding
SUPPORTS
OpenAI
PageIndex exports agent tools directly into OpenAI agents
Source finding
BENCHMARKED AGAINST
FinanceBench
PageIndex claims leading accuracy on FinanceBench.
Source finding
PARTNERS WITH
OpenAI Agents SDK
PageIndex now supports Agentic Vectorless RAG with OpenAI Agents SDK
Source finding
RELEASED
VectifyAI
VectifyAI built PageIndex with hierarchical tree indexes
Source finding
RELEASED
VectifyAI
PageIndex achieved 98.7% accuracy on FinanceBench via Mafin 2.5 without vector databases.
Source finding
COMPETES WITH
ChatPDF
PageIndex replaces ChatPDF-style single-document tools with folder-level retrieval.
Source finding
COMPETES WITH
Perplexity
PageIndex competes with Perplexity as a multi-document reasoning system.
Source finding
BENCHMARKED AGAINST
FinanceBench
PageIndex achieves 98.7% accuracy on FinanceBench with vectorless reasoning-based RAG without embeddings or chunking
Source finding
BENCHMARKED AGAINST
FinanceBench
PageIndex scores 98.7% on FinanceBench without embeddings or chunking
Source finding
BENCHMARKED AGAINST
Firecrawl Developer Index
PageIndex and Firecrawl Developer Index both ship with published retrieval accuracy metrics.
Source finding
SUPPORTS
Anthropic
PageIndex exposes agent bindings for Anthropic tools.
Source finding

Source trail

Bank Statement Converter Latent Space Product Hunt v0.2.11 openclaude yvgude/lean-ctx PageIndex VoltAgent's collection Shanraisshan's repo Paul Bakaus's project 3.0 release candidate GitHub

Graph sources

entity graphfindings textkg edgeskg entitiesnewsletter issues

Fetching from the wire…