When Detroit SMB owners say "AI chatbot," they usually mean three different things — and they only need one of them.
This is not a criticism. The terminology is genuinely confusing, and vendors exploit the ambiguity to sell products that do not match the actual need.
This post clears it up: what the three types of AI chat tools actually are, which one fits your business, what it costs to build, and the four mistakes that kill most deployments.
The Three Types of AI Chat Tools
Type 1: The FAQ Assistant
An FAQ assistant is an AI trained on your specific documents — your pricing page, service descriptions, FAQ, and policies — that answers customer questions via a chat widget on your website.
A well-built FAQ assistant:
- Answers your 15–20 most common customer questions accurately, 24/7
- Escalates anything uncertain to a human via email, phone, or live chat
- Never answers a question it was not trained on
- Reduces inbound email and phone volume by 30–40% in the first 90 days
Right for: service businesses, professional practices, retail, and any business where customer questions are predictable and repetitive.
Not right for: businesses where every customer interaction is unique and high-stakes — a custom fabrication shop where every quote requires specific technical knowledge, for example.
Type 2: The Booking and Intake Chatbot
A booking chatbot goes beyond answering questions to completing a transaction: scheduling an appointment, submitting an intake form, or qualifying a lead. The conversation ends with something booked, a form submitted, or a structured lead handed to your sales team.
Right for: service businesses with predictable booking flows — dental practices, HVAC scheduling, consulting intake, repair services.
The key difference from a voice agent: text versus voice. A booking chatbot lives in a web chat widget. A voice agent handles phone calls. Many businesses benefit from both — web chat for visitors, voice agent for callers.
Type 3: The AI-Powered Knowledge Agent
A knowledge agent connects to your internal systems — CRM, project tracker, order management — and retrieves or acts on real data. A customer asks "where is my order?" and the agent pulls the live status directly from your system.
This requires backend integration and more complex build work. For SMBs, it is the right choice only after FAQ and booking chatbots are already deployed and delivering value.
Real Use Cases for Detroit-Area SMBs
Professional services firms in Troy, Birmingham, Bloomfield Hills
Initial consultation scheduling, service Q&A, and basic intake — handled via website chat before the prospect ever calls. A 10-person law firm in Troy deployed an FAQ + booking chatbot and eliminated 60% of their inbound phone volume for new prospect inquiries in the first quarter.
Home services contractors in Macomb County
Service area Q&A, appointment booking for non-emergency jobs, and after-hours lead capture. Visitors who arrive at 9pm get immediate, accurate responses instead of a "contact us" form that goes nowhere until Monday morning.
Dental and medical practices in Oakland County
New patient FAQ, insurance Q&A, and appointment booking. Front desk staff stops fielding the same 15 questions all day and focuses on in-office patient care. Evening and weekend inquiries get handled without staff involvement.
What It Costs to Build a Chatbot for Your Business
| Type | Build Time | Build Cost | Monthly Platform |
|---|---|---|---|
| FAQ assistant | 8–20 hrs | $1,500–$4,000 | $30–$100 |
| Booking chatbot | 15–35 hrs | $3,000–$8,000 | $50–$150 |
| Knowledge agent (with backend integration) | 30–80 hrs | $8,000–$25,000 | $100–$400 |
These assume a production-ready deployment with proper escalation logic, testing, and documentation.
Budget vendors quoting $500 for a "full AI chatbot" are selling a template with your logo on it — fine for demos, not for customer-facing use where wrong answers have consequences.
The Four Most Common Chatbot Mistakes
1. Hallucinating answers.
An untrained or poorly configured chatbot will confidently invent information — pricing that does not exist, policies that are wrong, services you do not offer. Every chatbot must be constrained to answer only from its training data, with a hard fallback for anything it is not certain about. This is not optional.
2. No escalation path.
A chatbot without a "talk to a person" option traps frustrated customers. Always build in a clear, easy way out — email, phone, or live chat handoff. The escalation path should be visible, not buried.
3. Training on bad source material.
Your chatbot is only as accurate as what you feed it. An outdated pricing page, an FAQ with inconsistencies, service descriptions that no longer reflect your current offerings — these get amplified at scale. Audit your source documents before training.
4. Deploying and forgetting.
Customer questions change. Policies update. Services get added or removed. A chatbot with no maintenance plan goes stale within a quarter. Budget 2–4 hours per month to review conversations, identify new question patterns, and update training data.
Chatbot vs. Voice Agent: When You Need Both
For many Detroit-area service businesses, the right answer is both — a chatbot for web visitors and a voice agent for phone callers.
The two channels handle different customer behaviors. Web visitors are often in research mode — comparing options, checking pricing, deciding whether to contact you. A chatbot converts that research behavior into a booked appointment or a qualified lead.
Phone callers are typically in decision mode — they have already decided they need help, and they want someone to answer. A voice agent ensures every decision-mode caller gets an immediate, professional response.
Together, they cover the full inbound surface. Separately, each one leaves a gap.
Next Step
If you are not sure which type fits your business, the $297 AI Readiness Audit includes a chatbot and voice agent recommendation as part of the workflow assessment — mapping your customer contact volume, question types, and escalation requirements before you commit to any build.
For businesses ready to deploy, the AI Implementation Sprint covers chatbot builds as a primary use case — scoped, built, tested, and handed off in four weeks.
