Qian Lin Web
AI Agents & Automation

Automate the customer questions and repetitive work your team handles every day.

We build AI agents that answer common questions, capture leads, search approved business knowledge and hand conversations to a person when needed.

CRM, API, database and multi-system automation projects are quoted separately.

Customer questionConcept workflow · No client claim

Do you deliver outside the UK?

  1. 01Understand the request
  2. 02Search approved knowledge
  3. 03Check rules and confidence
  4. 04Prepare an answer
AnswerHuman handoff

Productised package

AI Customer Service Agent

$299

A focused website chat assistant for common questions and lead capture, configured from approved FAQ or website content.

Discuss the $299 package

Included

  • Website chat widget
  • Answers based on approved FAQ or website content
  • Lead capture
  • Human handoff
  • Basic branding
  • One language
  • One consolidated revision round
  • Basic deployment and testing

Not included

  • WhatsApp Business API
  • CRM integration
  • Order lookup
  • Database access
  • Custom API or tool calling
  • Voice agent
  • Multiple languages
  • Advanced RAG
  • Multi-agent workflows

Projects that need system access, multiple languages or more advanced retrieval and automation remain a Custom quote.

Practical AI tasks

What can an AI agent handle?

The useful starting point is a specific task with approved information, clear limits and a person who owns exceptions.

01

Customer questions

Answer common questions from your website, FAQ and business material you approve.

02

Lead capture

Collect contact details, requirements and budget information, then prepare a structured lead.

03

Knowledge search

Find relevant information in approved company knowledge, documents and product material.

04

Human handoff

Escalate when the agent cannot answer reliably or a question needs personal attention.

05

Workflow automation

Connect forms, email, CRM, documents and internal systems in a separately scoped workflow.

Interactive workflow

See what happens between a question and an answer.

Scroll through the steps or switch to the lead example. The confidence check determines whether the agent answers or pauses for a person.

Customer question

Concept workflow · No client claim

Incoming

Do you deliver outside the UK?

  1. 01Understand the request
  2. 02Search approved knowledge
  3. 03Check rules and confidence
  4. 04Prepare an answer
  5. 05Answer or request help
Outcome
AnswerHuman handoff

Automation with control

Human review stays where mistakes carry a cost.

The workflow can pause before customer messages, financial actions, record deletion or other sensitive steps. Operators can see the input, reason, output and failure state.

  • 01Approval before sensitive actions
  • 02Fallback queue for low-confidence results
  • 03Logs for inputs, outputs and errors
  • 04Clear owner for recovery and changes

Example use cases

Five useful places to start.

Choose a use case to see the basic flow. These examples explain possible system behaviour and do not represent client work.

Answer routine questions from approved support material and escalate uncertain requests.

  1. 01Request arrives
  2. 02Find approved answer
  3. 03Check confidence
  4. 04Answer or hand off
Concept workflow · No client claim

Advanced AI agents

Need more than a website chatbot?

A custom project can connect business systems, retrieve controlled data and run multi-step workflows with explicit permissions and recovery paths.

Custom Quote

Discuss a Custom Automation Project
  • 01CRM
  • 02Email
  • 03Website forms
  • 04Internal APIs
  • 05Databases
  • 06Documents
  • 07Messaging systems
  • 08Business software

Working process

From business workflow to a testable AI agent.

Each stage produces something the team can review before access or automation expands.

01

Discover the task

Identify the trigger, people, systems, current effort and the outcome worth improving.

02

Map the workflow

Record inputs, decisions, exceptions, approvals and the system that owns each field.

03

Prototype the critical path

Test the uncertain model or integration step with representative examples before full implementation.

04

Build the workflow

Connect the agreed systems and add validation, permissions, retries and operator controls.

05

Test edge cases

Exercise incomplete inputs, low-confidence output, service outages and manual recovery.

06

Launch and improve

Release with logs and ownership in place, then review real exceptions before expanding scope.

Before you start

Questions about the agent and its limits.

What content can the $299 agent use?

The standard package uses approved FAQ or website content for one language. Document collections, advanced retrieval and live system data require a custom quote.

Does the agent replace customer support staff?

No. It handles agreed routine questions and captures leads. Uncertain, sensitive or unsupported requests can be handed to a person.

How are advanced agents priced?

Advanced AI agents and automation are quoted after reviewing the workflow, integrations, data access, testing and operating requirements.

Your next website

Have a website in mind?

Let’s talk about what it needs to do.
Share your brief. We’ll review the scope and reply in writing.