AI Automation & Customer Response
What Should Be in an AI Receptionist Knowledge Base?
An AI receptionist knowledge base is only as useful as the information and operating rules behind it. A natural-sounding voice cannot compensate for outdated prices, vague service descriptions, or missing instructions for difficult questions.
The knowledge base should function like an operating manual for the front desk: what the business does, what it does not do, what the agent may say, and when a human must take over.
Knowledge Base Step 1: Business Identity
Start with the basics:
- official business name;
- public-facing brand name;
- website;
- phone numbers;
- business hours;
- holiday or after-hours rules;
- physical location or service-area model;
- approved greeting.
This reduces the risk of the AI describing the company inconsistently.
2. Service List
Define every service the receptionist may discuss. For each service, include:
- service name;
- plain-language description;
- who it is for;
- important exclusions;
- geographic limits;
- pricing approach;
- next step.
Do not include future services that the business cannot currently fulfill unless the system is explicitly instructed to treat them as unavailable.
3. Pricing Rules
If pricing is public, include the approved ranges. If projects require custom quotes, say so clearly.
Examples:
- “Website Design & Launch projects are generally $495–$2,995+ depending on scope.”
- “Exact pricing is provided after a consultation.”
- “Third-party software fees are separate unless the proposal says otherwise.”
Do not let the model invent discounts or quote custom work outside approved rules.
4. Service Areas
For local businesses, list the areas genuinely served. Include rules for nearby locations that may be reviewed individually.
The agent should know what to do when a caller is outside the normal area:
- politely explain the standard service area;
- collect the request for human review;
- avoid promising service that has not been approved.
5. Frequently Asked Questions
Build FAQs from real customer questions, not just marketing copy. Useful examples:
- How long does the process take?
- Do you serve my city?
- What does the service include?
- Do I need to provide access or materials?
- What happens after I book?
- Do you offer ongoing support?
Write direct answers that the AI can reuse accurately.
6. Lead Qualification Questions
Define the minimum information needed to determine the next step. For example:
- name;
- contact information;
- service requested;
- location;
- timeline;
- project or business type.
If the business needs additional details for one service, make those questions conditional instead of asking every caller everything.
7. Appointment and Scheduling Rules
If the receptionist can book appointments, document:
- appointment types;
- duration;
- available days and hours;
- buffer times;
- required lead information;
- rescheduling rules;
- cancellation rules;
- who can be booked;
- situations that require human approval.
The AI should never create a booking outside the real calendar rules.
8. Transfer and Escalation Rules
This may be the most important section. Tell the agent exactly when to stop handling the conversation and involve a human.
Examples:
- caller is angry or threatening;
- legal complaint;
- billing dispute;
- high-value custom project;
- emergency situation;
- question is not in the knowledge base;
- caller explicitly requests a human;
- AI confidence is low.
Include the correct phone number, department, or follow-up action for each escalation type.
9. Things the AI Must Never Promise
Create explicit guardrails. Depending on the business, these might include:
- guaranteed SEO rankings;
- guaranteed revenue;
- custom discounts;
- legal or regulatory advice;
- project completion dates not approved by staff;
- availability that has not been checked;
- services outside the approved list.
“Do not guess” should be a real operating instruction, not an assumption.
10. Policies
Include customer-facing policies that callers may ask about:
- deposits;
- payment methods;
- cancellations;
- refunds;
- travel charges;
- service minimums;
- warranties;
- privacy or recording disclosures where applicable.
11. CRM and Handoff Fields
If the receptionist creates lead records, define the fields that should be passed downstream:
- name;
- phone;
- email;
- service;
- location;
- qualification notes;
- summary;
- urgency;
- appointment details;
- source.
Consistent field names make integrations more reliable.
12. Example Conversations
Provide examples of good interactions:
- new lead;
- existing customer;
- out-of-area caller;
- pricing question;
- appointment request;
- complaint;
- question the AI cannot answer.
Examples help define tone and expected behavior more concretely than abstract instructions alone.
13. Tone and Brand Voice
Specify whether the agent should sound warm, concise, formal, upbeat, technical, or consultative. Avoid making the tone so elaborate that it interferes with clarity.
A receptionist’s job is to help the customer move forward, not deliver a brand manifesto on every call.
14. Change Log and Ownership
Assign responsibility for maintaining the knowledge base. Track meaningful updates such as:
- new service;
- price change;
- new phone number;
- holiday hours;
- service-area change;
- policy update;
- new booking rule.
An outdated knowledge base can produce confidently wrong answers.
How Often Should the Knowledge Base Be Reviewed?
Review it whenever the business changes and on a regular operating cadence. High-change businesses may need monthly review. Stable companies may need less frequent formal review but should still update immediately when pricing, hours, policies, or services change.
Test the Knowledge Base Before Launch
Run realistic scenarios and deliberately ask questions the AI should not answer. Test:
- normal FAQs;
- ambiguous requests;
- wrong service areas;
- pricing exceptions;
- angry callers;
- requests for human help;
- questions outside scope;
- background noise or unclear speech.
The Bottom Line on an AI Receptionist Knowledge Base
A good AI receptionist knowledge base is a controlled source of business truth plus operating rules. It should make the agent useful without giving it permission to improvise important business decisions.
For the broader implementation process, see AI Automation for Small Businesses and Nacluv Tech’s AI Automation Systems service.