What is this course about?
AI support tools priced per conversation or per resolution look cheap in a quiet month and expensive in your busiest one. This course models those pricing shapes against usage-based and owned options on your own volumes, then builds the support agent itself: which questions it answers, grounding on your policies, customer data protection and honest measurement.
Who is this course for?
- Small business owners and managers
- Office and customer service managers
- Bookkeepers and part-time finance staff
What do I need before starting?
- Your monthly customer message and call volumes for the last year, or a best estimate
- Any AI support quotes you have received
What will I be able to do afterwards?
- Classify AI support quotes by pricing shape
- Model per-conversation cost across your busiest and quietest months
- Estimate usage-based cost per resolved question
- Ground a support agent on your own policies and prices
- Measure resolution honestly rather than deflection
What does each module cover?
How is AI customer support priced?
30 minPer seat, per conversation, per resolution, usage-based and owned, and what each assumes about your volume.
Objectives
- Classify quotes by pricing shape
- Find the unit each shape charges for
- Spot pricing that prevents comparison
Topics
Activity. Classify the support quotes you have by pricing shape.
What does per-conversation pricing cost in your busy season?
30 minSeasonal volume multiplied by a per-conversation fee, month by month.
Objectives
- Model monthly cost from your volumes
- Find your most expensive month
- Model growth
Topics
Activity. Build the twelve-month per-conversation bill from your own volumes.
What does a resolved question cost on usage pricing?
30 minCost per resolved question from message length, model choice and volume.
Objectives
- Estimate usage per conversation
- Compare model choices
- Compute cost per resolved question
Topics
Activity. Compute cost per resolved question for two model choices.
Which questions should the agent answer, and which come to you?
30 minA safe set of questions to automate and a clean hand-off for everything else.
Objectives
- Sort past questions into automate and escalate
- Write the hand-off message
- Set business-hours and after-hours behavior
Topics
Activity. Sort fifty real customer questions into automate and escalate.
How do you ground the agent on your own policies and prices?
30 minYour returns policy, prices and hours as the agent's only sources, kept current.
Objectives
- Collect the documents the agent may use
- Test answers against them
- Set an update routine
Topics
Activity. Load your policies and test the agent on twenty questions.
How do you protect customer data in support conversations?
30 minCollect less, keep it for less time and control who can see it.
Objectives
- Limit what the agent collects
- Set retention for transcripts
- Control staff access
Topics
Activity. Write the support data policy in one page.
How do you measure resolution honestly?
30 minResolved questions, not deflected ones: recontact rate and customer feedback.
Objectives
- Define a resolved question
- Track recontacts
- Collect customer feedback
Topics
Activity. Define resolution for your business and the three numbers you will track.
Building your support cost model and launch plan
30 minThe workshop: the pricing comparison and a thirty-day launch plan.
Objectives
- Complete the pricing comparison
- Choose a pricing shape
- Write the thirty-day launch plan
Topics
Activity. Complete the model and the launch plan.
What is the capstone project?
Support pricing comparison and thirty-day launch plan
Model AI support cost for your business under per-conversation, per-resolution, usage-based and owned pricing on your real volumes, then plan a grounded support agent with escalation, data protection and honest measurement.
Deliverable: A twelve-month cost comparison and a one-page launch plan.
How are learners assessed?
- Cost model uses the business's real monthly volumes, including the busiest month
- Agent tested against twenty real customer questions
- Resolution defined as resolved, not deflected
What ships with the course?
Facilitator guide
Session-by-session running order, discussion prompts, and the questions that reliably derail a room.
Learner workbook
Exercises, checklists, and the templates each module's activity produces.
Hands-on lab environment
A sandboxed ibl.ai deployment so exercises run against real agents, not screenshots.
Assessment bank
Scenario questions and rubric criteria mapped to each stated learning outcome.
Source bibliography
Every primary regulation and standard cited on this page, linked and dated.
Which AI agents does this course use?
The hands-on modules run against agents already deployable on the ibl.ai platform for small business.
Where does the course material come from?
Every module is grounded in primary sources β the regulation, standard, or research itself, not a summary of it. Each was resolved at authoring time.
- AI Index Report
Stanford HAI
Inference cost trends underpinning usage-based modeling.
- Protecting Personal Information: A Guide for Business
Federal Trade Commission
Data minimization and retention practices for module 6.
- Small Business Cybersecurity Corner
NIST
Access control and security basics for support data.
- Business Blog
Federal Trade Commission
FTC guidance on AI claims and customer communications.
- Small Business Administration
SBA
Small business operations and planning resources.
Delivery notes
Binding guidance for anyone preparing and delivering this course.
- Have learners bring real quotes and real volumes. A model built on vendor examples always favors the vendor.
- Do not quote vendor list prices from memory. Use dated quotes the learners bring, because support pricing changes often.
- Module 7 must separate resolution from deflection; a closed chat is not a solved problem.
- Coordinate with SB-2 on support quality and SB-10 on the overall AI budget.
Why run AI training on a platform you own?
You own the course, not a licence to it
Course content, learner data, and the platform run inside your perimeter β you own all the code and the data.
Model-agnostic delivery
Run the course's AI components on any LLM β Claude, GPT, Llama, Gemini, Command β and switch anytime.
No per-seat training licences
Usage-based or self-hosted, so cost tracks actual use rather than headcount.
Deploy anywhere
Cloud, private VPC, on-premise, or fully air-gapped β including for cohorts that cannot use public AI tools.
Frequently asked questions
What does the AI Customer Support Without Per-Conversation Fees course cover?
AI support tools priced per conversation or per resolution look cheap in a quiet month and expensive in your busiest one. This course models those pricing shapes against usage-based and owned options on your own volumes, then builds the support agent itself: which questions it answers, grounding on your policies, customer data protection and honest measurement. It runs 4 hours across 8 modules, at foundational level, and closes with a capstone: Support pricing comparison and thirty-day launch plan.
Who should take AI Customer Support Without Per-Conversation Fees?
It is written for Small business owners and managers, Office and customer service managers, Bookkeepers and part-time finance staff. Prerequisites: Your monthly customer message and call volumes for the last year, or a best estimate; Any AI support quotes you have received.
Can we run this course on our own infrastructure?
Yes. ibl.ai is model-agnostic and deploy-anywhere β cloud, private VPC, on-premise, or fully air-gapped β and you own all the code and the data. Cohort data, submissions, and any material learners upload stay inside your perimeter, which matters for small business teams that cannot send work to a public AI tool.
How do we get access to AI Customer Support Without Per-Conversation Fees?
Request access and we will set it up for your cohort β hosted by ibl.ai, or running against your own deployment. Tell us the group size and timing you need, and whether it should run inside your own perimeter.
How much does AI training for small business cost on ibl.ai?
There is no per-seat pricing β you pay for usage or self-host and pay only for the infrastructure, so a 5,000-person rollout does not cost 5,000 licences. 1.6M+ users across 400+ organizations run the platform this way, including NVIDIA, MIT, and Syracuse University.