--- title: "OpenKB Review: Knowledge Compiler vs RAG" canonical: https://wavect.io/blog/openkb-review-vs-rag/ language: en description: "OpenKB review for CTOs: compare its LLM knowledge compiler with vector RAG on evidence, security, cost, deployment, permissions and enterprise readiness." image: "https://wavect.io/img/blog/headers/header_openkb-review-vs-rag.png" --- [**Back**](/blog/overview/) [![Kevin Riedl](/img/team/kevin.webp)](/team/kevin-riedl/) [Kevin Riedl](/team/kevin-riedl/) https://linkedin.com/in/wsdt 14 min read · 11 Aug 2026 Last reviewed August 11, 2026 [**Next**](/blog/open-knowledge-format-okf/) # OpenKB Review: Is a Knowledge Compiler Better Than RAG? TL;DR OpenKB is an Apache-2.0, developer-focused knowledge compiler that turns documents into an evolving, cross-linked Markdown wiki, then queries that compiled artifact with an LLM. It is compelling for curated research, due diligence and agent knowledge that must accumulate across sources. It is not a universal RAG replacement. A preregistered small-corpus study found stronger cross-paper connections and claim-level citation support from a compiled wiki, while vector RAG won single-fact lookup and query-token cost. Separate research found blind compilation can discard critical facts, making evaluation and refinement mandatory. OpenKB is a credible pilot platform, but its alpha package status, single optional bearer token and documented scaling roadmap mean enterprises still need permissions, SSO, tenant isolation, audit, evaluation and source-governance layers. Compare OpenKB, vector RAG and a hybrid against the same questions before choosing. **OpenKB is one of the clearest working implementations of the LLM knowledge-compiler idea, but it is not a drop-in replacement for enterprise RAG.** It reads source material, writes summaries and concept pages, updates cross-references, then lets an agent query the accumulated wiki. That changes the unit of work from “find chunks for this question” to “maintain a reusable body of knowledge.” The pattern came from [Andrej Karpathy's April 2026 LLM Wiki idea file](https://gist.github.com/karpathy/442a6bf555914893e9891c11519de94f). The commercial question is no longer whether the idea sounds good. It is whether OpenKB gives a team better verified answers than vector RAG after build cost, information loss, permissions and maintenance enter the picture. Our verdict after reviewing the project and current research on 11 August 2026: **pilot OpenKB for curated, synthesis-heavy knowledge work; keep vector RAG or a hybrid for large, fast-changing or permission-sensitive corpora.** This page owns the product-specific “OpenKB review” and “OpenKB vs RAG” intent. Our separate [Open Knowledge Format enterprise guide](/blog/open-knowledge-format-okf/) owns the portable format, while the [RAG production-readiness checklist](/blog/rag-production-readiness-checklist-eu/) owns retrieval, permissions and answer controls. ## What is OpenKB? **OpenKB is an open-source Python CLI and web workbench that compiles files and URLs into a persistent, interlinked Markdown wiki.** The [official OpenKB repository](https://github.com/VectifyAI/OpenKB) documents PDF, Word, PowerPoint, Excel, HTML, CSV, text, Markdown and URL input. The compilation pass creates or updates source summaries, concepts, entities, an index and a log. Query, chat, visualization, slide and agent-skill generators then consume that wiki. | Layer | What OpenKB does | What the buyer still owns | | --- | --- | --- | | Raw sources | Copies or converts selected files and URLs into a knowledge-base workspace | Source approval, classification, retention, deletion and immutable evidence | | Long-document index | Uses a hierarchical PageIndex tree for PDFs of 20 pages or more | OCR choice, parser acceptance, page-level verification and provider policy | | Compiled wiki | Writes summaries, concept pages, entity pages, links, index and activity log | Schema rules, factual review, contradiction handling and freshness ownership | | Generators | Queries, chats, visualizes and distils agent skills from the wiki | User permissions, application UX, evaluation, monitoring and incident response | Short documents are converted with MarkItDown and read as text. For long PDFs, OpenKB relies on [PageIndex's vectorless tree-index approach](https://github.com/VectifyAI/PageIndex): it builds a table-of-contents-like hierarchy and asks an LLM to navigate relevant sections. “No vector database” therefore does not mean “no retrieval.” It means retrieval uses document structure and reasoning instead of embedding similarity. ## How does a knowledge compiler differ from vector RAG? Vector RAG usually stores chunks and embeddings, retrieves a top-k set for each question, then assembles an answer. OpenKB performs more synthesis during ingestion. A new document can rewrite several concept pages before any user asks a question. | Decision factor | OpenKB knowledge compiler | Vector RAG | | --- | --- | --- | | Primary artifact | Readable, linked Markdown wiki | Source chunks, metadata and vector index | | Work at ingestion | High: summarize, merge, link and revise concepts | Usually lower: parse, chunk, enrich and embed | | Work at query time | Navigate compiled pages and source indexes | Embed the query, retrieve chunks and synthesize | | Cross-source synthesis | Precomputed and accumulated in concept pages | Reconstructed per query unless an additional summary or graph layer exists | | Single-fact lookup | Can fail if compilation omitted the fact | Strong when the relevant chunk is retrievable and permissions permit it | | Human inspection | Strong: pages, links and Git diffs are directly readable | Depends on retrieval tooling and index observability | | Freshness | Requires safe recompilation and review of changed pages | Requires source sync, re-embedding and stale-index controls | | Access control | Not solved by Markdown compilation | Can apply document and chunk ACL filters, if built correctly | | Best first fit | Curated research, due diligence, standards, product or domain knowledge | Large operational corpora, support search, frequently changing and per-user content | The useful comparison is not “old RAG versus new magic.” OpenKB itself combines a compiled wiki with reasoning-based retrieval. A mature system can also retrieve over the compiled pages, raw sources or both. The architecture is a spectrum: raw chunk retrieval, hierarchical retrieval, compiled summaries, explicit graphs and cached full context can coexist. ## Is OpenKB free, and what does it really cost? The software uses Apache 2.0, so there is no OpenKB licence fee for commercial use. The current [PyPI package record](https://pypi.org/project/openkb/) listed version 0.4.4, released 10 July 2026, and classified the project as alpha when we checked. That is a maturity signal, not a criticism. It tells procurement to pin versions, test upgrades and budget ownership around a fast-moving dependency. Total cost has at least six parts: 1. **Compilation tokens.** Every source can trigger summary, concept, entity and cross-reference work. A single ingest may update many pages. 2. **Retrieval and answer tokens.** Query cost remains. Vectorless does not mean model-free or zero-token. 3. **Document processing.** Local parsing covers common inputs; complex OCR and optional PageIndex Cloud introduce another provider and cost path. 4. **Human review.** Someone must inspect factual compression, contradictions, citations and sensitive output. 5. **Enterprise controls.** SSO, role-based access, tenant boundaries, audit, secrets, backups and deletion are product work. 6. **Evaluation and operations.** Golden questions, source-diff regression tests, latency, retries, model changes and incident handling need continued ownership. The right cost unit is not “price per embedded page.” Measure cost per accepted, source-backed answer or completed research task. Include compilation, query, retries and reviewer minutes. Our [AI agent cost-per-action framework](/blog/ai-agent-cost-per-action-2026/) gives that denominator. ## Does compiled knowledge beat RAG in independent research? **No architecture wins every knowledge task.** The most useful recent evidence is a [preregistered comparison of vector RAG and an LLM-compiled wiki](https://arxiv.org/abs/2605.18490) over 24 papers and 13 questions. The wiki was better at connecting findings across papers and its cited pages more often supported the exact claim. Vector RAG met the preregistered test for single-fact lookup and used far fewer query tokens. A decomposition-based RAG variant recovered most of the wiki's synthesis advantage at lower token cost, but not its claim-level citation advantage. That study is deliberately small and used LLM judges rather than humans. It does not prove that OpenKB beats a production RAG stack. It does give CTOs a better evaluation model: score synthesis structure, exact claim support and total cost separately. A single “answer quality” average hides the trade-off you need to buy. A second paper exposes the main failure mode. [WiCER evaluated wiki-memory compilation across 6,800 questions](https://arxiv.org/abs/2605.07068). Blind compilation scored far below raw full context and showed a 53% to 60% catastrophic failure rate in the reported setup because it discarded critical facts. An iterative evaluate-and-refine loop recovered much of the lost quality. The lesson is direct: compilation needs diagnostic questions and repair passes, not a one-shot summary prompt. | Claim | Evidence supports | Evidence does not support | | --- | --- | --- | | Compiled wikis help synthesis | Promising advantage on cross-paper connections in a small preregistered study | A universal advantage across corpora, models and workloads | | Compiled wikis improve citations | Better claim-to-cited-page support in that study | Automatic factual correctness after summaries rewrite sources | | RAG is obsolete | Nothing in the studies supports this | RAG remained strong for lookup and query cost | | Compilation can run unattended | Iterative evaluation can repair information loss | Blind compilation was safe enough for critical knowledge | ## What does Google OKF change for OpenKB? Portability is one of the strongest parts of the approach. Google Cloud introduced the [Open Knowledge Format v0.1](https://cloud.google.com/blog/products/data-analytics/how-the-open-knowledge-format-can-improve-data-sharing) in June 2026. It standardizes a small interoperability surface around Markdown, YAML frontmatter, links, index files and logs. OpenKB says its wiki pages are OKF-ready. That matters because a failed pilot can still leave readable, versioned artifacts. A team can inspect them in an editor, move them through Git or feed them to another search and agent system. It does not guarantee semantic correctness, perfect interoperability or continued conformance as the v0.1 format evolves. Validate the generated bundle in CI and keep raw sources separate from compiled pages. ## Is OpenKB enterprise-ready? **OpenKB is ready for a controlled developer or research pilot. Its public control surface is not a complete enterprise knowledge platform.** The current workbench and REST API make it easier to upload, compile, query, chat, lint and recompile. The documented authentication model remains intentionally local-first. The [official REST API guide](https://github.com/VectifyAI/OpenKB/blob/main/examples/rest-api/README.md) says authentication is off by default and can be enabled with one bearer token. It explicitly warns that binding to a non-loopback host without that token leaves every reachable knowledge base open. A single shared token can protect a pilot endpoint. It is not the same as SSO, user identity, role-based access, document-level authorization or tenant isolation. | Production concern | OpenKB gives you | Enterprise work still required | | --- | --- | --- | | Authentication | Optional bearer token for the API | SSO, lifecycle, service identities and short-lived credentials | | Authorization | Knowledge bases selected by name | User, group, tenant, source and field-level permission enforcement | | Auditability | Wiki log, source files and Git-friendly output | Identity-linked access logs, admin actions, model calls and incident evidence | | Data protection | Local files and configurable model providers | Classification, encryption, residency, retention, deletion and backup policy | | Quality | Linting, source summaries and cited query output | Golden sets, adversarial tests, factual review and change regression | | Scale | File-based wiki and long-PDF indexing | Capacity tests, concurrency, database strategy, queues and recovery | For sensitive company sources, retrieval-time authorization is the hard boundary. If a user may read one policy but not another, compiling both into the same concept page can leak restricted facts before the query layer even filters anything. Build permission-preserving compilation domains or keep those sources out of the compiler. Our [permissions-first RAG architecture](/blog/rag-permissions-sharepoint-confluence-drive/) explains the same rule for SharePoint, Confluence and Drive. ## When should you choose OpenKB, vector RAG or a hybrid? | Your dominant need | Recommended starting point | Reason | | --- | --- | --- | | Connect findings across a curated research set | OpenKB pilot | Persistent concept pages make synthesis visible and reusable | | Build a human-readable agent knowledge repository | OpenKB or OKF-native workflow | Markdown, links and Git create an inspectable artifact | | Search millions of frequently changing records | RAG or search platform | Incremental indexing and narrow retrieval are the core workload | | Enforce per-user source permissions | Permission-aware RAG first | Authorization must constrain retrieval before synthesis | | Answer both lookup and cross-source synthesis questions | Hybrid | Retrieve raw evidence for facts and compiled pages for relationships | | One small static handbook | Long context or ordinary search | A compiler and vector stack may both add unnecessary maintenance | Do not start with the technology. Start with a labelled question set. If most questions are “What is the current refund limit in policy X?”, retrieval and permissions dominate. If they are “How did our assumptions about market Y change across twelve reports?”, persistent synthesis becomes more valuable. If both matter, run both paths and route by question type. ## A ten-day OpenKB pilot plan 1. **Choose one owned domain.** Use 20 to 50 approved sources with one accountable subject-matter owner. Do not begin with the whole company drive. 2. **Preserve the evidence layer.** Hash raw sources, record dates and keep compiled pages in a separate directory. Never let the wiki silently replace originals. 3. **Create 30 blind questions.** Include single facts, multi-source synthesis, contradictions, missing answers, permission traps and recently changed facts. 4. **Freeze three baselines.** Compare current search or manual work, a simple vector RAG implementation and OpenKB on the same sources and model budget. 5. **Score claim support.** Review whether each important sentence is supported by the cited source, not only whether the answer sounds complete. 6. **Test information loss.** Recompile after adding a contradictory or updated source. Check what changed, what vanished and whether the old claim remains reachable. 7. **Attack the boundary.** Test prompt injection inside a document, unauthorised source mixing, malicious links, huge files, parser failures and a model-provider outage. 8. **Price the full loop.** Count ingest and query tokens, p50 and p95 latency, retries, review minutes, wrong answers and recovery work. 9. **Set a scale gate.** Proceed only if verified answer quality or reviewer time improves enough to fund permissions, operations and maintenance. 10. **Keep the exit reversible.** Export the wiki, schema, evaluation set and source manifest so another system can consume them. Wavect's [AI enablement service](/services/ai-enablement/) can build this comparison on your infrastructure instead of declaring a winner from a demo. The [Twinsoft AI case study](/case-studies/twinsoft-ai/) shows our approach to traceable AI output and reviewer control. For the commercial model, compare [AI enablement with generic AI consulting](/compare/ai-enablement-vs-generic-ai-consultancy/) before deciding whether you need implementation or a strategy document. ## Frequently asked questions ### What is OpenKB? OpenKB is an open-source Python knowledge compiler. It converts files and URLs into a persistent, cross-linked Markdown wiki, then provides query, chat, visualization and agent-skill generators over that wiki. ### Is OpenKB free for commercial use? Yes. OpenKB uses the Apache 2.0 licence. There is no licence fee, but model calls, document processing, hosting, security, evaluation, integration and human review still create total cost. ### Does OpenKB need a vector database? No. OpenKB uses full-text conversion for short documents and PageIndex tree indexing for long PDFs. Its retrieval still uses LLM reasoning and document structure, so vectorless does not mean retrieval-free or cost-free. ### Can OpenKB replace RAG? Sometimes for a curated, synthesis-heavy corpus, but not universally. Vector RAG remains a strong starting point for precise lookup, large changing corpora and permission-aware retrieval. A hybrid can query compiled concept pages and raw evidence together. ### Is OpenKB ready for enterprise deployment? It is credible for a controlled pilot. A production enterprise deployment still needs SSO, role and tenant authorization, source-level permissions, audit logs, secrets management, backups, retention, deletion, capacity testing and continuous evaluation. ### Does OpenKB work with Codex? The project includes an agent skill and documents a manual symlink setup for Codex. The skill reads the compiled wiki. Teams should still review the skill, pin the repository version and protect the wiki as internal source material. ### How is OpenKB related to Open Knowledge Format? OpenKB says its generated wiki pages are OKF-ready. OKF is the portable Markdown and YAML format; OpenKB is one producer and query tool. The format does not provide compilation quality, retrieval, permissions or governance by itself. ## Research boundary *Status checked 11 August 2026. This is an independent architecture and buyer review based on public project documentation and research, not a sponsored post, a penetration test or a hands-on benchmark of a private corpus. OpenKB, its package, API and roadmap can change quickly. Pin the version and verify current documentation before procurement.* ## Final thoughts OpenKB makes an important infrastructure shift concrete: knowledge can become a maintained product instead of temporary query context. Its readable wiki, long-document path and agent integrations make it a strong candidate for research and other synthesis-heavy work. The risks are equally concrete. Compilation can delete facts, query cost does not automatically fall, and a local-first bearer token is not enterprise authorization. Run a bounded comparison against vector RAG, score exact claim support, attack the permission boundary and keep the source layer immutable. Choose OpenKB only when accumulated synthesis creates measurable value that pays for those controls. ## You may also like.. [**Open Knowledge Format: The Enterprise Guide** Understand the portable Markdown format behind agent-ready knowledge and the governance it does not provide.](/blog/open-knowledge-format-okf/) [**AI Enablement vs Generic AI Consulting** Compare a working implementation on your infrastructure with a strategy-only engagement.](/compare/ai-enablement-vs-generic-ai-consultancy/) Models and infrastructure ## Continue through this cluster [Start with the cornerstone**Self-Hosting LLMs in the EU: When Open Weights Actually Pay Off**](/blog/self-hosting-llms-eu-cost/) - [Does Claude Watermark Text? The 2026 API Answer](/blog/claude-text-watermark-api-2026/) - [Unsloth Desktop Review: A Private Local AI Workstation?](/blog/unsloth-desktop-local-ai-workstation-review/) - [NeMo Switchyard 0.2: Agent Model Routing Without Training?](/blog/nemo-switchyard-model-router/) - [Firecrawl AnyDoc Review: 14 Formats to Markdown](/blog/firecrawl-anydoc-review/) - [Muse Glimmer 30B: Is Meta's Local Agent Model Production-Ready?](/blog/muse-glimmer-30b-local-agent-guide/) Inbox, without the noise ## Follow the work that matters to you Get a short email when we publish something new. Follow the whole blog or only the problems you care about. [**Back**](/blog/overview/) [![Kevin Riedl](/img/team/kevin.webp)](/team/kevin-riedl/) [Kevin Riedl](/team/kevin-riedl/) https://linkedin.com/in/wsdt 14 min read · 11 Aug 2026 Last reviewed August 11, 2026 [**Next**](/blog/open-knowledge-format-okf/) New posts by email × × Get new posts by email A short email when we publish. Free, no tracking. ## Structured Data ```json { "@context": "https://schema.org", "@graph": [ { "@id": "https://wavect.io/#organization", "@type": [ "Organization", "ProfessionalService", "LocalBusiness" ], "employee": [ { "@id": "https://wavect.io/team/kevin-riedl/#person", "@type": "Person", "jobTitle": "Managing Director", "name": "Kevin Riedl", "url": "https://wavect.io/team/kevin-riedl/", "worksFor": { "@id": "https://wavect.io/#organization", "@type": [ "Organization", "ProfessionalService", "LocalBusiness" ] } }, { "@id": "https://wavect.io/team/christof-jori/#person", "@type": "Person", "jobTitle": "Managing Director", "name": "Christof Jori", "url": "https://wavect.io/team/christof-jori/", "worksFor": { "@id": "https://wavect.io/#organization", "@type": [ "Organization", "ProfessionalService", "LocalBusiness" ] } } ], "founder": [ { "@id": "https://wavect.io/team/kevin-riedl/#person", "@type": "Person", "jobTitle": "Managing Director", "name": "Kevin Riedl", "url": "https://wavect.io/team/kevin-riedl/", "worksFor": { "@id": "https://wavect.io/#organization", "@type": [ "Organization", "ProfessionalService", "LocalBusiness" ] } }, { "@id": "https://wavect.io/team/christof-jori/#person", "@type": "Person", "jobTitle": "Managing Director", "name": "Christof Jori", "url": "https://wavect.io/team/christof-jori/", "worksFor": { "@id": "https://wavect.io/#organization", "@type": [ "Organization", "ProfessionalService", "LocalBusiness" ] } } ], "legalRepresentative": [ { "@id": "https://wavect.io/team/kevin-riedl/#person", "@type": "Person", "jobTitle": "Managing Director", "name": "Kevin Riedl", "url": "https://wavect.io/team/kevin-riedl/", "worksFor": { "@id": "https://wavect.io/#organization", "@type": [ "Organization", "ProfessionalService", "LocalBusiness" ] } }, { "@id": "https://wavect.io/team/christof-jori/#person", "@type": "Person", "jobTitle": "Managing Director", "name": "Christof Jori", "url": "https://wavect.io/team/christof-jori/", "worksFor": { "@id": "https://wavect.io/#organization", "@type": [ "Organization", "ProfessionalService", "LocalBusiness" ] } } ], "name": "Wavect GmbH", "subjectOf": { "@id": "https://wavect.io/verified-claims.json#dataset", "@type": "Dataset", "creator": { "@id": "https://wavect.io/#organization", "@type": [ "Organization", "ProfessionalService", "LocalBusiness" ] }, "description": "A machine-readable registry of quantitative and qualitative claims published by Wavect, with review dates, localized page appearances and public third-party citations where available.", "inLanguage": "en", "isAccessibleForFree": true, "license": "https://creativecommons.org/licenses/by/4.0/", "name": "Wavect verified publication claims", "url": "https://wavect.io/verified-claims.json" }, "url": "https://wavect.io/" }, { "@id": "https://wavect.io/team/kevin-riedl/#person", "@type": "Person", "jobTitle": "Managing Director", "name": "Kevin Riedl", "sameAs": [ "https://www.wikidata.org/wiki/Q139796365", "https://www.linkedin.com/in/wsdt", "https://github.com/wsdt" ], "url": "https://wavect.io/team/kevin-riedl/", "worksFor": { "@id": "https://wavect.io/#organization", "@type": [ "Organization", "ProfessionalService", "LocalBusiness" ] } }, { "@id": "https://wavect.io/team/christof-jori/#person", "@type": "Person", "jobTitle": "Managing Director", "name": "Christof Jori", "sameAs": [ "https://www.wikidata.org/wiki/Q139796367", "https://www.linkedin.com/in/jocr77/", "https://github.com/jo-chris" ], "url": "https://wavect.io/team/christof-jori/", "worksFor": { "@id": "https://wavect.io/#organization", "@type": [ "Organization", "ProfessionalService", "LocalBusiness" ] } }, { "@id": "https://wavect.io/#website", "@type": "WebSite", "inLanguage": [ "en", "de", "es", "zh" ], "name": "Wavect", "potentialAction": { "@type": "SearchAction", "query-input": "required name=search_term_string", "target": { "@type": "EntryPoint", "urlTemplate": "https://wavect.io/search/?q={search_term_string}" } }, "publisher": { "@id": "https://wavect.io/#organization", "@type": [ "Organization", "ProfessionalService", "LocalBusiness" ] }, "url": "https://wavect.io/" }, { "@id": "https://wavect.io/blog/openkb-review-vs-rag/#webpage", "@type": "WebPage", "dateModified": "2026-08-11", "inLanguage": "en", "isPartOf": { "@id": "https://wavect.io/#website", "@type": "WebSite" }, "lastReviewed": "2026-08-11", "url": "https://wavect.io/blog/openkb-review-vs-rag/" } ] } ``` ```json { "@context": "https://schema.org", "@type": "BlogPosting", "abstract": "OpenKB is an Apache-2.0, developer-focused knowledge compiler that turns documents into an evolving, cross-linked Markdown wiki, then queries that compiled artifact with an LLM. It is compelling for curated research, due diligence and agent knowledge that must accumulate across sources. It is not a universal RAG replacement. A preregistered small-corpus study found stronger cross-paper connections and claim-level citation support from a compiled wiki, while vector RAG won single-fact lookup and query-token cost. Separate research found blind compilation can discard critical facts, making evaluation and refinement mandatory. OpenKB is a credible pilot platform, but its alpha package status, single optional bearer token and documented scaling roadmap mean enterprises still need permissions, SSO, tenant isolation, audit, evaluation and source-governance layers. Compare OpenKB, vector RAG and a hybrid against the same questions before choosing.", "articleBody": " Blog overview/AI and agents/Models and infrastructure OpenKB Review: Is a Knowledge Compiler Better Than RAG? TL;DR OpenKB is an Apache-2.0, developer-focused knowledge compiler that turns documents into an evolving, cross-linked Markdown wiki, then queries that compiled artifact with an LLM. It is compelling for curated research, due diligence and agent knowledge that must accumulate across sources. It is not a universal RAG replacement. A preregistered small-corpus study found stronger cross-paper connections and claim-level citation support from a compiled wiki, while vector RAG won single-fact lookup and query-token cost. Separate research found blind compilation can discard critical facts, making evaluation and refinement mandatory. OpenKB is a credible pilot platform, but its alpha package status, single optional bearer token and documented scaling roadmap mean enterprises still need permissions, SSO, tenant isolation, audit, evaluation and source-governance layers. Compare OpenKB, vector RAG and a hybrid against the same questions before choosing. OpenKB is one of the clearest working implementations of the LLM knowledge-compiler idea, but it is not a drop-in replacement for enterprise RAG. It reads source material, writes summaries and concept pages, updates cross-references, then lets an agent query the accumulated wiki. That changes the unit of work from “find chunks for this question” to “maintain a reusable body of knowledge.” The pattern came from Andrej Karpathy's April 2026 LLM Wiki idea file. The commercial question is no longer whether the idea sounds good. It is whether OpenKB gives a team better verified answers than vector RAG after build cost, information loss, permissions and maintenance enter the picture. Our verdict after reviewing the project and current research on 11 August 2026: pilot OpenKB for curated, synthesis-heavy knowledge work; keep vector RAG or a hybrid for large, fast-changing or permission-sensitive corpora. This page owns the product-specific “OpenKB review” and “OpenKB vs RAG” intent. Our separate Open Knowledge Format enterprise guide owns the portable format, while the RAG production-readiness checklist owns retrieval, permissions and answer controls. What is OpenKB? OpenKB is an open-source Python CLI and web workbench that compiles files and URLs into a persistent, interlinked Markdown wiki. The official OpenKB repository documents PDF, Word, PowerPoint, Excel, HTML, CSV, text, Markdown and URL input. The compilation pass creates or updates source summaries, concepts, entities, an index and a log. Query, chat, visualization, slide and agent-skill generators then consume that wiki. LayerWhat OpenKB doesWhat the buyer still owns Raw sourcesCopies or converts selected files and URLs into a knowledge-base workspaceSource approval, classification, retention, deletion and immutable evidence Long-document indexUses a hierarchical PageIndex tree for PDFs of 20 pages or moreOCR choice, parser acceptance, page-level verification and provider policy Compiled wikiWrites summaries, concept pages, entity pages, links, index and activity logSchema rules, factual review, contradiction handling and freshness ownership GeneratorsQueries, chats, visualizes and distils agent skills from the wikiUser permissions, application UX, evaluation, monitoring and incident response Short documents are converted with MarkItDown and read as text. For long PDFs, OpenKB relies on PageIndex's vectorless tree-index approach: it builds a table-of-contents-like hierarchy and asks an LLM to navigate relevant sections. “No vector database” therefore does not mean “no retrieval.” It means retrieval uses document structure and reasoning instead of embedding similarity. How does a knowledge compiler differ from vector RAG? Vector RAG usually stores chunks and embeddings, retrieves a top-k set for each question, then assembles an answer. OpenKB performs more synthesis during ingestion. A new document can rewrite several concept pages before any user asks a question. Decision factorOpenKB knowledge compilerVector RAG Primary artifactReadable, linked Markdown wikiSource chunks, metadata and vector index Work at ingestionHigh: summarize, merge, link and revise conceptsUsually lower: parse, chunk, enrich and embed Work at query timeNavigate compiled pages and source indexesEmbed the query, retrieve chunks and synthesize Cross-source synthesisPrecomputed and accumulated in concept pagesReconstructed per query unless an additional summary or graph layer exists Single-fact lookupCan fail if compilation omitted the factStrong when the relevant chunk is retrievable and permissions permit it Human inspectionStrong: pages, links and Git diffs are directly readableDepends on retrieval tooling and index observability FreshnessRequires safe recompilation and review of changed pagesRequires source sync, re-embedding and stale-index controls Access controlNot solved by Markdown compilationCan apply document and chunk", "articleSection": "Engineering", "author": { "@id": "https://wavect.io/team/kevin-riedl/#person", "@type": "Person", "name": "Kevin Riedl", "sameAs": [ "https://www.wikidata.org/wiki/Q139796365", "https://www.linkedin.com/in/wsdt", "https://github.com/wsdt" ], "url": "https://wavect.io/team/kevin-riedl/" }, "citation": [ { "@type": "WebPage", "name": "Andrej Karpathy's April 2026 LLM Wiki idea file", "url": "https://gist.github.com/karpathy/442a6bf555914893e9891c11519de94f" }, { "@type": "WebPage", "name": "official OpenKB repository", "url": "https://github.com/VectifyAI/OpenKB" }, { "@type": "WebPage", "name": "PageIndex's vectorless tree-index approach", "url": "https://github.com/VectifyAI/PageIndex" }, { "@type": "WebPage", "name": "PyPI package record", "url": "https://pypi.org/project/openkb/" }, { "@type": "WebPage", "name": "preregistered comparison of vector RAG and an LLM-compiled wiki", "url": "https://arxiv.org/abs/2605.18490" }, { "@type": "WebPage", "name": "WiCER evaluated wiki-memory compilation across 6,800 questions", "url": "https://arxiv.org/abs/2605.07068" }, { "@type": "WebPage", "name": "Open Knowledge Format v0.1", "url": "https://cloud.google.com/blog/products/data-analytics/how-the-open-knowledge-format-can-improve-data-sharing" }, { "@type": "WebPage", "name": "official REST API guide", "url": "https://github.com/VectifyAI/OpenKB/blob/main/examples/rest-api/README.md" } ], "dateModified": "2026-08-11", "datePublished": "2026-08-11", "description": "OpenKB is an Apache-2.0, developer-focused knowledge compiler that turns documents into an evolving, cross-linked Markdown wiki, then queries that compiled artifact with an LLM. It is compelling for curated research, due diligence and agent knowledge that must accumulate across sources. It is not a universal RAG replacement. A preregistered small-corpus study found stronger cross-paper connections and claim-level citation support from a compiled wiki, while vector RAG won single-fact lookup and query-token cost. Separate research found blind compilation can discard critical facts, making evaluation and refinement mandatory. OpenKB is a credible pilot platform, but its alpha package status, single optional bearer token and documented scaling roadmap mean enterprises still need permissions, SSO, tenant isolation, audit, evaluation and source-governance layers. Compare OpenKB, vector RAG and a hybrid against the same questions before choosing.", "headline": "OpenKB Review: Knowledge Compiler vs RAG", "image": "https://wavect.io/img/blog/headers/header_openkb-review-vs-rag.svg", "inLanguage": "en", "keywords": "OpenKB, AI Knowledge Base", "mainEntityOfPage": { "@id": "https://wavect.io/blog/openkb-review-vs-rag/", "@type": "WebPage" }, "publisher": { "@id": "https://wavect.io/#organization", "@type": [ "Organization", "ProfessionalService", "LocalBusiness" ] }, "url": "https://wavect.io/blog/openkb-review-vs-rag/", "wordCount": 2795 } ``` ```json { "@context": "https://schema.org", "@type": "BreadcrumbList", "itemListElement": [ { "@type": "ListItem", "item": "https://wavect.io/", "name": "Home", "position": 1 }, { "@type": "ListItem", "item": "https://wavect.io/blog/overview/", "name": "Blog overview", "position": 2 }, { "@type": "ListItem", "item": "https://wavect.io/blog/topics/ai-agents/", "name": "AI and agents", "position": 3 }, { "@type": "ListItem", "item": "https://wavect.io/blog/clusters/models-infrastructure/", "name": "Models and infrastructure", "position": 4 }, { "@type": "ListItem", "item": "https://wavect.io/blog/openkb-review-vs-rag/", "name": "OpenKB Review: Knowledge Compiler vs RAG | ", "position": 5 } ] } ``` ```json { "@context": "https://schema.org", "@type": "FAQPage", "mainEntity": [ { "@type": "Question", "acceptedAnswer": { "@type": "Answer", "text": "OpenKB is an open-source Python knowledge compiler. It converts files and URLs into a persistent, cross-linked Markdown wiki, then provides query, chat, visualization and agent-skill generators over that wiki." }, "name": "What is OpenKB?" }, { "@type": "Question", "acceptedAnswer": { "@type": "Answer", "text": "Yes. OpenKB uses the Apache 2.0 licence. There is no licence fee, but model calls, document processing, hosting, security, evaluation, integration and human review still create total cost." }, "name": "Is OpenKB free for commercial use?" }, { "@type": "Question", "acceptedAnswer": { "@type": "Answer", "text": "No. OpenKB uses full-text conversion for short documents and PageIndex tree indexing for long PDFs. Its retrieval still uses LLM reasoning and document structure, so vectorless does not mean retrieval-free or cost-free." }, "name": "Does OpenKB need a vector database?" }, { "@type": "Question", "acceptedAnswer": { "@type": "Answer", "text": "Sometimes for a curated, synthesis-heavy corpus, but not universally. Vector RAG remains a strong starting point for precise lookup, large changing corpora and permission-aware retrieval. A hybrid can query compiled concept pages and raw evidence together." }, "name": "Can OpenKB replace RAG?" }, { "@type": "Question", "acceptedAnswer": { "@type": "Answer", "text": "It is credible for a controlled pilot. A production enterprise deployment still needs SSO, role and tenant authorization, source-level permissions, audit logs, secrets management, backups, retention, deletion, capacity testing and continuous evaluation." }, "name": "Is OpenKB ready for enterprise deployment?" }, { "@type": "Question", "acceptedAnswer": { "@type": "Answer", "text": "The project includes an agent skill and documents a manual symlink setup for Codex. The skill reads the compiled wiki. Teams should still review the skill, pin the repository version and protect the wiki as internal source material." }, "name": "Does OpenKB work with Codex?" }, { "@type": "Question", "acceptedAnswer": { "@type": "Answer", "text": "OpenKB says its generated wiki pages are OKF-ready. OKF is the portable Markdown and YAML format; OpenKB is one producer and query tool. The format does not provide compilation quality, retrieval, permissions or governance by itself." }, "name": "How is OpenKB related to Open Knowledge Format?" } ] } ```