"How much does an AI chatbot cost?" is one of the most common questions we get asked — and one of the hardest to answer honestly, because the truthful answer is: it depends enormously. A £30/month SaaS tool and a £150,000 custom enterprise deployment are both "AI chatbots." Here's how to understand the difference.
Tier 1 — Off-the-shelf platforms: £30–£500/month
The lowest-cost entry point. Platforms like Intercom, Tidio, Zendesk AI, and Drift offer pre-built AI chat widgets that you can add to your website in hours. You configure your knowledge base, set a tone, and go live.
Good for: businesses that want to handle FAQ-style enquiries and have relatively standard support flows. Setup is minimal and iteration is fast.
Limitations: limited ability to connect to your own systems (CRM, databases, live order data), constrained customisation of conversation logic, and you're dependent on the platform's model and feature roadmap. Typically not suitable for complex workflows or sensitive data that can't leave your infrastructure.
The cheapest solution is only cheap if it actually solves your problem. A poor fit costs more in the long run than building it right.
Tier 2 — No-code custom builds: £2,000–£15,000 + ~£200–£600/month
The middle ground. Using platforms like Voiceflow, Botpress, or Stack AI, an agency or technically capable in-house team can build a significantly more customised chatbot — including integrations with your existing systems, custom conversation flows, and branded UI — without writing bespoke backend code from scratch.
The one-off build cost typically covers: discovery and requirements, conversation design, knowledge base structuring, system integrations, testing, and deployment. Ongoing costs cover platform licences and AI model usage (API calls).
Good for: small to mid-size businesses that need more than a basic FAQ bot but don't have complex security or compliance requirements.
Tier 3 — Custom-built AI agent: £15,000–£80,000+
A fully custom solution built by a development team — using foundation models (Claude, GPT-4, Gemini) combined with your own infrastructure. This tier is appropriate when you need:
- Deep integration with proprietary systems, databases, or real-time data feeds.
- Fine-tuned behaviour that no off-the-shelf platform can match.
- Data sovereignty — the model and data must remain within your own infrastructure.
- Multi-step agentic workflows, not just conversation.
- Regulated industries (financial services, healthcare, legal) where compliance requirements are strict.
Running costs vary widely based on usage volume and the models used, but typically range from £500–£5,000/month for a well-optimised deployment.
Enterprise and multi-agent systems: £80,000–£500,000+
Large-scale AI deployments — multiple interconnected agents, custom model fine-tuning, enterprise security and compliance frameworks, integration across dozens of internal systems — fall into a different category entirely. These are transformation programmes as much as software projects.
What drives cost up or down
| Factor | Why it affects the price |
|---|---|
| Number of integrations | Each connected system (CRM, ERP, inventory, booking, etc.) adds complexity and testing time. |
| Quality of existing data | A clean, well-structured knowledge base significantly reduces build time. Messy, fragmented documentation does the opposite. |
| Conversation complexity | A bot that answers FAQs is orders of magnitude simpler than one that executes multi-step processes with conditional logic. |
| Compliance requirements | GDPR, FCA, HIPAA, and similar frameworks add architecture, testing, and documentation overhead. |
| Ongoing maintenance | Models need retraining as your products and policies change. Build this cost into your total. |
The question to ask first
Before committing to any budget, answer this: what is the value of the work you're automating? If your support team handles 500 enquiries per week at a cost of £8 per enquiry, and an AI agent can handle 70% of them, you're looking at £145,600 of annual value. A £25,000 build cost plus £500/month running cost pays back in under three months.
Build the business case first, then size the investment accordingly. The businesses that overbuild (spending £80k on something £15k would solve) and underbuild (saving £3k upfront only to replace it six months later) both lose. Getting the scope right from the start is the most valuable thing any good AI partner can help you with.