AI assistants and search decision
Wholana
A single developer can build a useful personal research + semantic-search workflow (ingest, embeddings, UI, assistant) but cannot cheaply reproduce Wholana's proprietary Egyptian-TikTok corpus, daily sweep, MCP integration, and multi-seat collaboration at product parity.
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
- Ingest TikTok posts for a niche, index them with semantic search, present a filterable feed with decoded craft metadata, let the user save items into collections and write scripts alongside referenced items, and run an assistant query over the indexed corpus.
What it still won’t have
- The daily, large-scale Egyptian-TikTok sweep and breadth of the proprietary corpus
- Built-in MCP server integration for ChatGPT / Claude as a workspace feature
- Multi-seat workspace collaboration, live co-editing and seat-based billing
- WhatsApp intake and mobile-first sharing flows out of the box
- Scale reliability and continuous outlier-scoring on a growing corpus
What remains hard
- Proprietary data
The filterable feed over the Egyptian TikTok corpus
- Proprietary data
Search, filters, and craft analysis all work on Egyptian-dialect content.
First-year cost
Keep paying
Paying is—cheaper in year one.
On cash alone, building overtakes the subscription at 26 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 self-hosted content-research service in Next.js (React) + Postgres (+ pgvector) with a Python ingestion service (FastAPI) and worker (RQ/Celery). In scope: (1) a crawler that accepts TikTok video URLs and stores metadata and captions, (2) embedding generation using OpenAI embeddings and a vector index, (3) a React UI with a filterable feed and semantic search showing per-video decoded fields (hook, format, subject), (4) save-to-collection and simple script editor that pins references, (5) an assistant endpoint that queries the vector index and calls an LLM (OpenAI) to synthesize answers and returns source video links. Out of scope: large-scale automated daily sweeps, WhatsApp intake, multi-tenant billing, live co-editing, and advanced outlier-scoring. Include error handling, retries, basic unit tests for ingestion, and a docker-compose deployment with environment-variable config for API keys.
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
- Hard moats found in the evidence-3
- Evidence score64
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 productWholana — Home
- official pricingWholana — Pricing
- official docsWholana — Features
- open sourcearc53/DocsGPT
- open sourcedzhng/deep-research
Integrity checks
What held up, and what did not.
