How simple keyword matching and predefined responses create convincing "AI-like" experiences at a fraction of the cost
The demo chatbot uses three simple techniques to create an AI-like experience
When you type "pain" or "headache", the system scans for these specific words and routes you to the Symptoms section. It's not understanding context—just pattern matching like Ctrl+F.
Typing indicators with animated dots, deliberate response delays (0.8-1.4 seconds), and conversational flow create the impression of intelligence and "thinking" when it's just a timer.
Each response has predefined buttons that navigate to the next node. Like a "Choose Your Own Adventure" book—every path is pre-written, not generated on the fly.
| Feature | Decision Tree (This Demo) | Real AI (GPT, Claude) |
|---|---|---|
| Cost per query | £0.00 | £0.05 - £0.50 |
| Response time | Instant (fake delay added) | 1-3 seconds |
| Handles unexpected questions | ✗ Breaks immediately | ✓ Adapts naturally |
| Context memory | ✗ Forgets everything | ✓ Remembers conversation |
| Natural language understanding | ✗ Keywords only | ✓ Understands meaning |
| Compliance & control | ✓ Every word is auditable | ✗ Can hallucinate |
| Setup time | 1-3 days | 2-8 weeks (training, testing) |
| Maintenance | Edit configuration (no coding) | Ongoing retraining, monitoring |
| Works offline | ✓ Fully local | ✗ Requires API |
| Handles complex reasoning | ✗ No reasoning capability | ✓ Multi-step problem solving |
80% of charity/organization FAQs fall into predictable categories
Symptoms → Treatment → Support → Emergency contacts. Highly structured, repeatable questions.
How to donate → Tax receipts → Monthly giving → Volunteer info. Clear pathways.
Eligibility checks → Required documents → How to apply → Where to submit. Compliance-critical.
Troubleshooting → Warranty → Returns → Manuals. Limited, known issues.
Requires understanding context, nuance, and multi-factor analysis. Needs real AI.
"Tell me about recent developments in..." needs synthesis and reasoning.
"Based on my symptoms AND medical history..." requires context awareness.
"Help me design a campaign for..." needs ideation and adaptation.
Most users don't know they're talking to a decision tree for the first 2-3 interactions.
The typing dots, response delays, and conversational tone are enough to suspend disbelief. By the time they ask an unexpected question and it breaks, they've often already found what they needed.
For charities with tight budgets: if 80% of your queries are "How do I donate?", "Where are you located?", "What services do you offer?"—you don't need AI. You need a well-designed decision tree.
The Real Question: Are you buying a Ferrari (AI) to drive to the corner shop, when a forklift (decision tree) would do the job perfectly?
See for yourself how convincing the decision tree chatbot looks within our dummy health support web-site
Try the Demo →Six ready-built templates for healthcare, charity, education, professional services, restaturant and a basic default
Industry Templates →