Plan for $100–$1,500 a month to keep an AI chatbot healthy after launch: platform or API fees ($50–$500), content and knowledge upkeep (2–8 hours a month of someone's time, or $150–$800 on a retainer), and monitoring/tuning. A chatbot is not a crock-pot — you can't set it and forget it, because your products, policies and prices change, and the bot answers from what it was given. Here's the full post-launch budget, by tier.
Maintenance cost by chatbot tier
| Tier | Platform/API | Content upkeep | Monitoring & tuning | Realistic monthly total |
|---|---|---|---|---|
| Configured off-the-shelf | $50–$500 | 1–2 hrs/mo | Vendor dashboard | $100–$600 |
| Custom build | $100–$800 | 2–5 hrs/mo | Monthly review | $300–$1,200 |
| Advanced (RAG/agentic) | $500–$3,000+ | 4–8 hrs/mo | Weekly + eval suite | $800–$4,000 |
(The build-cost side of these tiers is covered in what an AI chatbot costs to build.)
Where the money actually goes
- Platform and API usage. LLM-powered bots pay per conversation under the hood. Usage grows with success — a bot that deflects twice the tickets costs more in API calls, which is the good kind of bill.
- Knowledge upkeep. Every price change, policy update and new product has to reach the bot's knowledge source, or it confidently gives last quarter's answer. This is the most-skipped item and the most common reason bots decay.
- Conversation review. Someone should read a sample of transcripts monthly: what did it get wrong, where did users give up, what's being asked that it can't answer? Your bot's admin dashboard makes this a 30-minute job, not an afternoon.
- Model and platform changes. Providers deprecate models and change APIs. A custom bot needs a developer-hour or two when that happens; platforms handle it for you inside the subscription.
What skipping maintenance costs
A neglected bot doesn't fail loudly — it decays. Answer accuracy drifts as your business changes, users learn to distrust it, and containment (the % of conversations it resolves) slides quietly. Six months of neglect typically shows up as "our chatbot doesn't work" and a rebuild quote, which costs more than the maintenance would have. It's the same pattern as the broader hidden costs of AI automation.
In-house vs retainer
If you have someone technical-ish with 2–4 hours a month, in-house works for tier one and two — the tasks are editorial more than technical. Go retainer when the bot is customer-facing revenue infrastructure or advanced tier: $300–$1,500/month buys transcript review, knowledge updates, tuning and a monthly report. The retainer vs project pricing breakdown covers how those engagements are structured, and agency pricing vs DIY covers who to buy them from.
Budgeting a chatbot end-to-end and want the honest total — build and run? Book a free consultation and we'll price your specific case, including whether you need one at all.
Frequently asked questions
How much does it cost to maintain a chatbot per month?
A configured off-the-shelf bot runs $100–$600/month all-in, a custom bot $300–$1,200, and an advanced RAG or agentic bot $800–$4,000+. The components are platform/API fees, knowledge and content upkeep, and monitoring/tuning time.
What happens if you don't maintain an AI chatbot?
It decays quietly rather than failing loudly: answers drift out of date as prices and policies change, users learn to distrust it, and resolution rates slide. Six months of neglect typically ends in a rebuild quote that costs more than the maintenance would have.
Can I maintain a chatbot in-house?
For simple and mid-tier bots, yes — the work is editorial more than technical: 2–5 hours a month updating knowledge and reviewing transcripts. Move to a retainer ($300–$1,500/month) when the bot is customer-facing revenue infrastructure or uses RAG/actions that need technical tuning.
Yash
Founder & Principal Consultant, Ynexgen
Yash leads Ynexgen, helping small and mid-sized businesses turn technology into a stronger foundation for growth — 7+ years across Salesforce CRM, websites, and AI adoption.
