Writing and content decision
Afforai
A small team or single competent developer can build a useful replacement MVP (search+fetch+LLM synthesis) in about a week; the vendor likely adds polish, hosting, and scale but no irreplaceable moats are evident from the provided page.
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.
What a replacement has to do
- Accept a research query → run web search and fetch source pages → extract and index source text/metadata → call an LLM to synthesize a cited report/summary and surface source citations
What it still won’t have
- proprietary, pre-built web-scale search/indexing
- polished UI/UX and hosted front-end
- any proprietary models or backend optimizations the vendor may run
- turnkey integrations and single-tenant hosting
What remains hard
- Product polish and ongoing maintenance
First-year cost
No published price
Afforai does not publish a price we could read, so there is nothing to compare against. What building costs is below.
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 self-hosted minimal research-and-write web app using Next.js for the frontend, FastAPI for the backend, Postgres for storage, and OpenAI (or Anthropic) for LLM calls. In scope: single-user auth, a search endpoint that queries Bing/Google Custom Search, page fetcher that extracts main text, storage of documents and metadata, a retrieval step (simple BM25 or vector search), and an LLM-driven synthesis endpoint that returns a report with inline numbered citations and a bibliography. Out of scope: multi-tenant billing, advanced UI polish, custom model training, large-scale crawling. Include error handling for failed fetches, rate-limit/backoff for search and LLM APIs, and unit tests for fetch, storage, retrieval, and synthesis endpoints.
How we checked
How the score was reached
- Build verdict base78
- An open-source build was found+5
- 2 cited sources+1
- Evidence score84
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 · 2
Every page the run actually retrieved.
- official productLogically | Research, Cite, and Write With AI
- open sourcezi-yue-1129/DATAGEN
Integrity checks
What held up, and what did not.
