Key takeaways
- A website FAQ assistant can launch in weeks; enterprise assistants take a few months.
- Integrations, security and evaluation drive cost more than the language model itself.
- Budget for running costs: model usage, hosting, monitoring and content updates.
Indicative cost ranges
Costs depend on scope. As a general guide for custom builds by an experienced partner:
| Type | Typical timeline | Indicative budget (USD) |
|---|---|---|
| Website FAQ assistant | 2–4 weeks | $8k – $20k |
| Knowledge assistant (RAG) | 4–8 weeks | $20k – $60k |
| Enterprise assistant with integrations and SSO | 8–16 weeks | $60k – $150k+ |
These are indicative ranges only; a short discovery call lets us give you a fixed or ranged estimate for your exact needs.
What drives the cost?
- Data preparation — cleaning, structuring and indexing your documents for retrieval.
- Integrations — CRM, help desk, e-commerce platform and single sign-on.
- Security and compliance — PII redaction, private deployments, HIPAA or SOC 2 requirements.
- Evaluation — test sets and automated checks for accuracy and safety.
- Channels — web, mobile, WhatsApp, Slack, Teams or voice.
Want a precise number for your chatbot?Send us your use case and get a fixed-scope estimate within 48 hours.
Get a free estimate What are the running costs?
Expect monthly costs for model usage (often a few cents per conversation), vector database hosting, monitoring and periodic content refreshes. Routing simple questions to smaller, cheaper models and reserving larger models for hard ones can reduce spend significantly.
How to keep it accurate
Ground every answer in your own sources with citations, set clear fallback rules (“I don’t know — let me connect you to a person”), and review conversation analytics every week. See how we approach this on our AI Solutions page.
