Support
StandardHuman handover
AI FAQ Agent
A bounded first-line agent that answers repeated questions from your approved knowledge, works out of hours and hands over cleanly to a person when a request falls outside its scope.
Request an architecture review AI readiness diagnosticIn short
An AI FAQ Agent answers the questions your team already answers dozens of times a week, using only knowledge you have approved. It covers first-line enquiries across your chosen channels, keeps answers consistent, works out of hours and classifies each request. Crucially, it hands over to a person the moment a question leaves its defined scope — complaints, legal or financial matters, anything sensitive. It is not a general chatbot that tries to answer everything; its value is a clear boundary and a reliable escalation path. The safest first step is a short review of your knowledge sources, escalation rules and channels.
Best for: Head of customer support · Operations manager · Business owner without 24/7 cover
Typical context: Businesses with repetitive first-line questions · Multilingual customer bases · Companies losing enquiries out of hours or in messengers
When leadership usually needs this
The same questions are answered over and over by hand.
Response time swings depending on who is online.
Different employees give different answers to the same question.
Knowledge is scattered across docs, chats and people’s heads.
Customers expect answers outside business hours.
Supporting several languages is hard to staff.
Escalation is inconsistent and things fall through.
Experts spend time on basic, repeatable questions.
There is no visibility into what customers keep asking.
What the agent actually does
- Answers a question only when it matches your approved knowledge, in scope.
- Stays silent or escalates when a question is outside its defined boundary.
- Classifies each request by topic for reporting and routing.
- Hands over to a human with the conversation context on sensitive or complex cases.
- Keeps answers consistent across channels and employees.
- Works out of hours and in multiple languages where configured.
- Logs unanswered questions so the knowledge base can be improved.
Inputs and outputs
Inputs and required data
Events / triggers
Incoming messages on connected channelsChat or form questionsOut-of-hours enquiries
Approved knowledge
Curated FAQ / help articlesProduct and policy documentsTone and answer templates
Business rules
Scope boundary (what it may answer)Escalation and handover rulesBusiness hours and languages
System access
Channel connections (web, messenger)Help-desk / inbox for handoverLeast-privilege access
Outputs and business actions
- An in-scope answer grounded in approved knowledge
- A human handover with context when out of scope
- A topic classification for each request
- A conversation summary for the support team
- A log of questions it could not answer (knowledge gaps)
- Consistent multilingual first-line responses
How the workflow works
1
Receive question
A message arrives on a connected channel.
2
Match to knowledge
The agent checks whether an approved answer covers it.
3
Answer in scope
If covered, it responds using approved content and tone.
4
Classify
It tags the request by topic for reporting.
5
Detect boundary
If out of scope or sensitive, it stops answering.
6
Escalate
It hands over to a person with the conversation context.
7
Log gaps
Unanswered questions are recorded to improve the knowledge base.
Human oversight and handover
Automated
- In-scope answers
- Topic classification
- Out-of-hours first response
- Gap logging
Requires human approval
- Publishing or changing the approved knowledge
- Widening the answer scope
- Any commitment, refund or contractual reply
Escalation
- Complaints, legal/financial questions, high-risk or emotional cases, and refund/contract requests are handed to a person.
Ownership: The support relationship and every non-routine decision stay with your team. The agent handles the repetitive first line and knows when to step back.
Systems and integrations
Commonly integrated
Website chatWhatsApp / TelegramEmail inboxHelp-desk tools
Possible via API / webhook
Knowledge-base platforms via APICRM for contact contextCustom channels via API / webhook
Requires technical assessment
Voice / telephonyRegulated-content domainsProprietary internal systems
Qualitative business impact
Faster, more consistent first-line answers
Coverage out of hours and across languages
Experts freed from repetitive questions
Cleaner escalation and fewer dropped enquiries
Visibility into recurring questions and gaps
Actual impact depends on process quality, data quality, integration scope, user adoption and governance.
When it is not suitable
There is no approved, maintainable knowledge base.
Answers change constantly or are highly case-specific.
Enquiry volume is very low.
You expect it to answer everything, including sensitive matters.
There is no human escalation path available.
Risks, constraints and governance
It answers confidently but wrongly (hallucination)
It answers only from approved knowledge and within a defined scope; outside it, it escalates rather than guesses.
Stale knowledge gives outdated answers
A clear owner and review cadence for the knowledge base; gap logging surfaces drift.
A sensitive case is mishandled
Explicit handover rules move complaints, legal, financial and emotional cases to a person immediately.
Customers feel stuck with a bot
A visible, easy path to a human is designed in from the start.
Alternatives and simpler options
A well-structured self-service FAQ page
Better when: Questions are stable and customers will read.
Help-desk canned responses / macros
Better when: A human is always available and just needs speed.
A rule-based chatbot with fixed flows
Better when: Questions fit a small, fixed decision tree.
AI FAQ Agent vs a generic chatbot
| Dimension | AI FAQ agent | Generic chatbot |
|---|---|---|
| Answer source | Approved knowledge only | Whatever the model produces |
| Out of scope | Escalates to a human | Often guesses |
| Consistency | Controlled tone and content | Variable |
| Escalation | Designed handover with context | Frequently missing |
| Best when | Real knowledge + clear boundary | Casual, low-stakes chat |
Implementation sequence
- Process discovery — top questions, channels, escalation needs.
- Knowledge audit — what is approved, current and maintainable.
- Architecture design — scope boundary and handover rules.
- Pilot scope — one channel and a defined topic set.
- Integration — connect channels and the handover inbox.
- Testing and approval rules — validate answers and escalation.
- Launch — go live on the pilot scope.
- Monitoring and improvement — close knowledge gaps from the logs.
What affects scope and price
Number and type of channelsSize and maturity of the knowledge baseNumber of languagesComplexity of scope and escalation rulesHelp-desk / CRM integrationVolume and monitoring needsTone and brand-voice work
Exact scope and price are defined after a short architecture review.
Frequently asked questions
No — that is the point. It answers only what your approved knowledge covers, within a defined scope, and escalates anything else to a person.
A generic chatbot generates answers freely and often guesses. The FAQ Agent is grounded in approved content and has a designed escalation path for out-of-scope and sensitive cases.
Only from the knowledge you approve — your FAQ, help articles and policy documents. It does not invent policy.
It stops and hands over to a person immediately, with the conversation context, per your escalation rules.
Yes, where configured — first-line consistency across languages is one of its strongest use cases.
There is no fixed price — scope depends on channels, knowledge maturity, languages and integrations. Exact scope and price follow a short architecture review.
Map your first-line answers
We start with a short review of your knowledge sources, escalation rules and channels. You get a clear scope and a safe first step — no full implementation is assumed before the review.
Request an architecture review AI readiness diagnostic