Recipes for ContextuAI Solo's local RAG Knowledge Base. Map a folder of PDFs, docs, and spreadsheets, and every agent, crew, and bot answers from your source of truth — fully on-device, no cloud required.
A Knowledge Base turns your files into retrievable context. Point it at a folder, and Solo indexes it locally so your AI can cite the exact passages that matter — instead of hallucinating.
Point a Knowledge Base at a directory of docs. Add files or re-index as your content changes — no manual copy-paste.
Retrieval pulls the most relevant passages into the model's context, so replies are anchored to your real documents.
Pair a KB with a local GGUF model and sensitive files never leave your machine — ideal for legal, HR, and finance.
Attach a KB to a single agent, a multi-agent crew, or an inbound support bot — one source of truth, many surfaces.
Create a Knowledge Base from a folder or uploaded files — indexed locally for retrieval.
Map a folder or upload PDFs, DOCX, Markdown, and CSV files into a new Knowledge Base.
Attach the KB to an agent, crew, or inbound channel bot so responses retrieve from it.
Chat, run a crew, or let a bot reply — every answer is grounded in your documents.
Each recipe names the scenario, the features to combine, and how to set it up.
Answer customer questions from your product docs — with a human approving every reply.
Features: Knowledge Base (product docs) + single-agent crew + Telegram or Discord (inbound) + Approvals
Setup: Map your /docs folder as a KB. Bind a single-agent crew to it. Connect Telegram or Discord inbound. Enable approvals so the AI drafts a doc-grounded reply and you approve before it sends.
Knowledge BaseInboundApprovalsCompare vendor contracts against your standards and flag risky clauses — without data leaving your machine.
Features: Knowledge Base (contract templates) + Local GGUF model + Crew (Sequential) + Contract Specialist, General Counsel agents + PDF output
Setup: Upload standard templates as a KB. Use a local model for confidentiality. Sequential crew: Contract Specialist compares clauses to templates → General Counsel flags risks and redlines. Export annotated PDF.
Local ModelPrivacyPDFLet new hires ask policy, benefits, and tooling questions and get instant, handbook-accurate answers.
Features: Knowledge Base (handbook, benefits, tool guides) + People Operations Manager agent + Telegram or Discord (inbound) + Approvals
Setup: Upload onboarding PDFs as a KB. Bind the People Operations Manager agent to it. Connect an inbound channel. Enable approvals for the first weeks, then disable once the bot proves reliable.
Knowledge BaseHROnboardingAnswer RFPs faster by pulling from past proposals, product docs, and technical specs.
Features: Knowledge Base (past RFPs + product docs) + Crew (Pipeline) + Solutions Architect, Technical Writer agents + PDF output
Setup: Map proposals and product-doc folders as KBs. Pipeline crew: Solutions Architect matches RFP questions to your capabilities via KB retrieval → Technical Writer formats the response. Export to PDF.
Knowledge BaseCrewPDFCheck your privacy policy against GDPR/CCPA text and flag non-compliant sections.
Features: Knowledge Base (policy + regulation text) + Local model + Crew (Sequential) + Privacy Officer, Compliance Officer agents + Markdown output
Setup: Upload your policy and the regulation text as KBs. Use a local model. Sequential crew: Privacy Officer reviews each section against requirements → Compliance Officer flags gaps and suggests language. Output to Markdown.
Local ModelCompliancePrivacyAnswer setup questions in your Discord using project docs, and escalate anything complex.
Features: Knowledge Base (project docs + FAQ) + Discord (inbound) + single-agent crew + @mention triggers + Approvals
Setup: Map your /docs directory as a KB. Bind a single-agent crew to it. Connect Discord inbound with @mention triggers so the bot only replies when tagged. Approve replies during the first week, then trust it in known channels.
DiscordTriggersInboundYes. The Knowledge Base is a local RAG store. You map folders or upload files from your machine and retrieval happens on-device. Paired with a local GGUF model, no document content leaves your computer.
Common business documents such as PDF, DOCX, Markdown, TXT, and CSV. Map a folder of docs or upload files, then agents and crews retrieve grounded context from them.
Yes. A Knowledge Base can be bound to a single agent, a multi-agent crew, or an inbound channel bot, so every response is grounded in your source documents.
A Knowledge Base scales past the context window. Instead of pasting everything, retrieval selects only the most relevant passages per question — so answers stay accurate even across hundreds of documents.
Download ContextuAI Solo for free, map a folder, and build a doc-grounded assistant in minutes.