AI assistants and search decision
Chatwith
A single developer can build a useful RAG chatbot and embeddable widget in about a week, but reproducing Chatwith’s full product (integrations catalogue, white‑label agency features, polished analytics and host-scale guardrails) is larger and better served by the hosted product or self-hosting mature prior-art projects.
Visit website↗Open-source builds that already do this
Every project below is open source and already does this job today. Fork one, self-host it, or take the parts you need - the build prompt further down assumes an empty file, and this is the shortcut past that. Licences differ; check the one on each card before you ship. All Chatwith alternatives, with the arithmetic →
What a replacement has to do
- Ingest site and files → index into a vector store → answer user queries via an LLM with RAG → surface answers in an embeddable website widget → collect conversations and basic analytics.
What it still won’t have
- Polished multi-tenant white-label dashboard and client portal
- Large catalogue of prebuilt integrations and no-code action connectors
- Usage-based billing, subscription management, and UX polish
- Hosted scale, SLA, and battle-tested abuse guardrails
- Built-in multilingual tuning, analytics charts, and automated daily retrain flows
What remains hard
- Product polish and ongoing maintenance
First-year cost
Keep paying
Paying is—cheaper in year one.
On cash alone, building overtakes the subscription at 6 seats.
Money you would actually spend
Time you would spend
—
What you would spend
What we assumed
The verdict above measures whether you could build it. This one is only about money.
Runnable build prompt
Build a minimal hosted RAG chatbot service using Next.js (React) frontend + Node/Express backend, Postgres with PGVector for vector storage, and OpenAI-compatible APIs for embeddings and completions. In scope: (1) a site scraper that extracts page text and splits into chunks, (2) file upload/parsing for PDF/DOCX/TXT and YouTube transcript ingestion, (3) embedding pipeline storing vectors in PGVector and a retriever+prompt assembly, (4) a simple embeddable chat widget (JS snippet) that proxies user messages to the backend, (5) an admin UI to view conversations, trigger re-index, and export logs via webhook. Out of scope: multi-tenant billing, a marketplace of prebuilt integrations, white-label client portal, and advanced analytics dashboards. Require error handling, input validation, test coverage for ingestion, retrieval, and end-to-end chat flows, and CI config to deploy to a single small cloud VM (e.g., DigitalOcean/Hetzner).
How we checked
How the score was reached
- Partly verdict base52
- An open-source build was found+5
- 5 cited sources+3
- Price verified on pricing page+3
- 3/3 assessment runs agreed+4
- Evidence score67
The base comes from the verdict. Everything under it is a check that either happened or did not, and each one is a fact frozen in this record rather than a judgement made at render time - so the same evidence always produces the same number.
How scoring works →Cited sources · 5
Every page the run actually retrieved.
- official productChatwith - Custom ChatGPT chatbot with your website & files
- official pricingPricing - Chatwith
- official productCustom chatbot for a real estate agency
- open sourcearc53/DocsGPT
- open sourcejohnbean393/Sidekick
Integrity checks
What held up, and what did not.
