Customer support decision
Whisperchat
A competent developer can build a useful site-trained AI support widget using open-source RAG and chat projects and hosted LLMs, but recreating WhisperChat's full polished product, analytics, Odoo app integration and managed scaling is larger and operationally heavier than a one-person replacement.
Visit website↗$49/mo
$588/yr
Read off the official pricing page.
$100one-off54 h to build
$100/mo6 h/mo upkeep
On cash alone, building overtakes the subscription at 3 seats.
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 Whisperchat alternatives, with the arithmetic →
What a replacement has to do
- User asks question in website widget → query vector DB of site content → generate grounded answer via LLM → display answer with confidence indicator → optionally capture lead and escalate to human (webhook/email).
What it still won’t have
- Proprietary model optimizations and 'full model access' tuning provided by vendor
- Polished multi-site management, UX polish, and brand-free iframe option
- Built-in paid analytics and recurring-question insights at vendor scale
- SLA, commercial support, and managed scaling for high-volume traffic
What remains hard
- Brand trust
First-year cost
Keep paying
Paying is—cheaper in year one.
On cash alone, building overtakes the subscription at 3 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 AI website support chatbot self-hosted service using Node.js (Express) backend, PostgreSQL + PGVector for vector storage, Python scripts for content crawling/processing, OpenAI (or compatibility layer) for embeddings and LLM calls, and a small React widget served as an embeddable iframe. In scope: website crawler + ingestion pipeline, embedding creation and vector store, a REST API endpoint that accepts user messages, k-nearest retrieval and prompt construction, LLM call for answer generation, confidence-score heuristic, lead-capture form that posts to a webhook/email, minimal analytics (question counts, recurring-question detection), and an embeddable chat widget with escalation UI. Out of scope: multi-tenant billing UI, advanced model fine-tuning, SLA-grade autoscaling, and a polished admin console. Include error handling, retries for API calls, input validation, and unit tests for core backend logic.
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
- Evidence score63
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 productWhisperChat product page
- official pricingWhisperChat pricing
- official docsWhisperChat API docs
- open sourcechatwoot/chatwoot
- open sourcepapercups-io/papercups
Integrity checks
What held up, and what did not.
