Writing and content decision
fastwrite.io
A competent developer can build a useful subset (Word add-in + PDF-backed retrieval + LLM prompts) in a few weeks, but reproducing fastwrite's claimed large scholarly database, polish, and evaluation features is expensive and time-consuming.
Visit website↗Not priced
No pricing page we fetched carried a figure, so there is nothing to compare against. The build side is still real.
$100one-off120 h to build
$200/mo6 h/mo upkeep
No published price to break even against.
No open-source build does this yet
Nothing published replaces this one, so a replacement starts from an empty file. Here is what it would have to cover.
What a replacement has to do
- Upload personal literature (PDFs) → parse and index PDFs for semantic search → provide inline AI autocomplete and citation suggestions inside Word via an add-in → fetch LLM completions from a hosted model API → insert citations and formatted references back into the document.
What it still won’t have
- Access to fastwrite's claimed database of “more than 30 Mio. scientific sources”
- Polished, native Windows/macOS desktop apps and one-click Word integration polish
- Built-in Deep Review evaluation workflow (9 university criteria)
- Existing user base, UI polish, and any proprietary model optimizations claimed
What remains hard
- Product polish and ongoing maintenance
First-year cost
No published price
fastwrite.io 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 minimal self-hosted academic writing assistant: stack: React frontend + Office Web Add-in for Word, Node.js + Express backend, PostgreSQL for metadata, vector store (FAISS or pgvector), and OpenAI-compatible LLM calls. Core features in scope: (1) user authentication, (2) upload PDFs with per-page text extraction and store page offsets, (3) generate embeddings and store in vector index, (4) semantic search endpoint returning top-k passages, (5) Word Add-in UI to request an autocomplete for the current selection that calls backend with context + retrieved passages, (6) citation insertion that maps passages to formatted reference + page number, (7) tests for PDF parsing, retrieval, and the Add-in flow, (8) error handling for upload failures, API timeouts, and malformed PDFs. Out of scope: training or hosting your own LLMs, building a 30M-source scholarly database, desktop native installers (macOS/Windows native apps), and automated plagiarism-detection. Require automated unit and integration tests, CI that runs the test suite, and basic docs to deploy on a single VPS.
How we checked
How the score was reached
- Partly verdict base52
- 5 cited sources+3
- 3/3 assessment runs agreed+4
- Evidence score59
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 productfastwrite.io - home
- official productfastwrite.io - features / home
- official productfastwrite.io - features list
- official productfastwrite.io - features list
- official productfastwrite.io - pricing block
Integrity checks
What held up, and what did not.
