In short
The problem and the solution
The problem
Service quality depends on who picks up. Policies live in old emails. When you add a chatbot or voice assistant, it inherits the same confusion at scale.
The solution
We capture your policies and your best people's answers, you approve them, and your team and assistants all draw on the same source with a person behind it.
What changes
Customers hear one answer, assistants stay inside approved content, and the business carries less risk from what its AI says.
A story you may recognize
The chatbot that promised a refund the airline did not offer
In November 2022, Jake Moffatt was booking a flight to a family funeral. He asked the airline's website chatbot about bereavement fares. The chatbot told him he could apply for the reduced fare after travel. The airline's own policy page said something different.
How the story ends, and what it means for you
When the refund was refused, the case went to a tribunal in British Columbia. The airline argued, remarkably, that the chatbot was responsible for its own words. In February 2024 the tribunal rejected that. A chatbot is part of the company's website, it said, and the company is responsible for all of it.
The damages were small, a few hundred dollars. The lesson was much bigger. Whatever answers your customers, a person or a machine, speaks for your business.
Klarna learned a related lesson. In early 2024 it said its AI assistant handled two-thirds of its customer service chats in its first month. In 2025 it began hiring people again. Its CEO said cost had been too dominant a factor, and the result was lower quality.
The fix in both cases looks the same: one approved source of truth, clear limits on what an assistant may answer, and a person ready for the moments that need judgment.
Source: Moffatt v. Air Canada, 2024 BCCRT 149, summarized by ABA Business Law Today; CX Dive and Entrepreneur on Klarna
What the research shows
You are not alone with this
AI support helps customers too
In the study of 5,179 support agents, AI assistance improved customer sentiment and reduced requests for managers to step in.
Source: Brynjolfsson, Li and Raymond, Generative AI at Work, NBER Working Paper 31161
Promises are hard to keep
Only 26% of employees believe their organization always delivers on its promises to customers.
Source: Gallup
Company case studies
Companies that faced the same problem
These businesses tackled the same problem in their own way. Each case study sets out what they did and where a Company Brain could take it next.
What two companies did about it, one case study each
Trailstone Insurance · Insurance agency
Trailstone moved its procedures into one searchable system. According to the vendor story, onboarding dropped from three to five days to one and a half days at most, while the agency grew to 68 people.
Read the case study
Experience Your Wealth · Financial planning
Jake Northrup built a private AI assistant from the firm's procedures and more than seven years of client conversations.
Read the case study
Your step-by-step plan
What you can do, starting this week
Each step assumes your ExValu Company Brain is in place: your documents and the know-how of your people, connected so your team can ask questions and get sourced answers. The time estimates are typical and depend on your business.
- 1
Collect the 50 questions customers ask most1 hour
Pull them from your inbox, phone notes and complaint log. Include the awkward ones about refunds, delays and exceptions.
- 2
Write one approved answer per questionTwo 60-minute sessions
Your best service people and you agree the answer. Your Company Brain stores it with an owner and a review date, so nobody quotes last year's terms.
- 3
Give the team the same source30 minutes
Staff ask the Company Brain during calls and emails. A team member types "Can a customer return an item after 30 days if it was a gift?" and sees the approved answer and when to involve a manager.
- 4
Connect your chat and voice assistantsHalf a day
Your website chat and phone assistant answer from the same approved content. Anything outside it goes to a person, with the conversation attached.
- 5
Set the escalation rules45 minutes
Decide which topics always go to a human: complaints, money above a set amount, health or safety, anything emotional. Your Company Brain enforces the list.
- 6
Review new questions every week20 minutes a week
Questions the Company Brain could not answer become new approved answers. Policy changes update the source once, and every channel follows.
What it does for the value of your business
What a buyer sees
Without a Company Brain
Each customer's experience depends on who they reach, and any AI channel repeats whatever inconsistencies exist. A buyer sees service quality tied to individuals and unmanaged risk from automated answers.
With a Company Brain
Every channel answers from one approved, dated source, with a clear handover to a person. A buyer sees consistent service that protects recurring revenue and a governed AI setup.
Strategic buyers look for service quality they can roll into their own customer base without surprises.
How you will know it is working: four numbers to track
Measure it
How you will know it is working
- Share of customer questions answered from approved content
- Repeat contacts on the same issue
- Complaints about wrong or conflicting information
- Escalations to a person, by topic
Privacy and trust: how your knowledge stays private
Privacy and trust
Customer data stays in your own systems, and the Company Brain answers from approved policy content. Your assistants say they are assistants, and customers can always reach a person.
Want to talk it through?
Answer a few short questions to see if the Company Brain fits, then pick a time for a 30-minute call with our team.
Where every fact on this page comes from
Sources
- Moffatt v. Air Canada, 2024 BCCRT 149, as summarized by ABA Business Law Today, February 2024
- CX Dive, Klarna changes its AI tune and again recruits humans, May 2025; Entrepreneur, Klarna CEO reverses course
- Brynjolfsson, Li and Raymond, Generative AI at Work, NBER Working Paper 31161
- Gallup, 6 Scary Numbers for Your Organization's C-Suite
Facts and figures are as reported in the sources above, checked by our team on October 11, 2026. Company-reported and vendor figures are described as such. To request a correction, write to [email protected].
ExValu - Human vision, AI output