Views · 7 daysTop 29%
LLM App Platforms & RAG
Platforms for building LLM applications: retrieval-augmented generation, agent memory, model routing and orchestration.
28 results in LLM App Platforms & RAG
Active Pro listings from this category appear first. Your sort applies within Pro and Free results.
1 views57,632 stars
TypebotCustomer SupportTypebot is a fair source chatbot builder: design flows from bubbles, inputs, logic and integration blocks, embed them on a site or WhatsApp number, run them over HTTP and collect answers as results. Cloud plans or self-hosting with Docker.CommercialSelf-hostedFreemiumViews · 7 daysTop 29%
ZolaAI Assistants & ChatZola is an Apache-2.0 web chat interface built on Next.js, the Vercel AI SDK and Supabase. It connects hosted model providers, bring-your-own API keys and local Ollama models, and runs on Vercel, Docker or your own Node.js server.Open sourceSelf-hostedViews · 7 daysNo views in 7 days
Shortlists for LLM App Platforms & RAG
Explore collections with tools relevant to this category, grouped by deployment, source model or workload.
Collaboration tools with APIs for developer teams →
4 matching tools from this category
Self-hosted IT tools →
27 matching tools from this category
IT tools with an API →
27 matching tools from this category
How to choose llm app platforms & rag tools
LLM app platforms supply the plumbing between a language model and a working application. Listings here build retrieval-augmented generation pipelines that fetch relevant documents before the model answers, store long-term memory as knowledge graphs, route each request to a cheaper or more capable model through one gateway, compress context to save tokens and offer visual builders for chat and agent workflows. Agent frameworks are code libraries for agent logic, and an MCP server exposes tools and data to an AI client through the Model Context Protocol, so both sit elsewhere.
Who uses these tools
- Developers building chatbots over company documents
- Platform teams standardising access to model providers
- AI engineers tuning retrieval quality and cost
Capabilities to compare
- Document ingestion, chunking and embeddings
- Vector, keyword and graph retrieval
- Model routing, fallbacks and usage quotas
- Visual workflow and agent builders
- Context compression and token tracking
Operating considerations
- Documents sent to external embedding providers
- Token costs multiply with retrieved context
- Project activity can stop or shift suddenly
- Access control must follow source document permissions
Selection checklist
- Build a small evaluation set of real questions and measure answer accuracy before and after adding retrieval.
- Check which model providers, embedding models and vector stores are supported without custom code.
- Confirm document permissions carry through retrieval so users only get answers from content they can access.
- Review the project's maintenance status and licence, since tools in this space change direction quickly.
What belongs in this category: A product belongs here when it helps developers assemble, ground, route or optimise LLM-powered applications through retrieval, memory, model gateways or workflow builders, rather than being an end-user chat assistant or a single agent.
LLM App Platforms & RAG on iThub in figures
iThub lists 28 tools in LLM App Platforms & RAG. 26 of 28 are open source, 27 of 28 can be self-hosted, 28 of 28 have a free, freemium or open-source pricing model and 27 of 28 offer an API.
- Tools listed
- 28
- Open source
- 26
- Self-hostable
- 27
- Free to start
- 28
- With an API
- 27
- Web app
- 24
- Desktop app
- 15
Typical use cases
- Answering employee questions from internal documentation
- Routing simple prompts to cheaper models automatically
- Giving agents persistent memory across sessions
- Prototyping a chat workflow without writing backend code
Trade-offs to model
- Visual builders speed up prototypes, while code-first libraries give finer control over retrieval, testing and deployment.
- Self-hosted platforms with local models keep data private, whereas hosted APIs deliver stronger models with less infrastructure.
Common mistakes
- Shipping a retrieval chatbot without an evaluation set, so quality regressions go unnoticed after changes.
- Indexing every document regardless of permissions and letting users retrieve content they should not see.
Decision resources
Continue the technical evaluation
Use the resources that match your decision stage. Profile data and linked sources were refreshed through November 11, 2026.
Comparisons
Dify vs Flowise →0 tools selectedCompare selected