Valentina Akpan: Founder of Rellatech, providing administrative and operations support to executives, founders, business owners and teams. Her background combines technical support, customer success, administration and operations.
The Starting Point
Support answers were kept in two FAQ documents maintained by different teams. Many questions appeared in both, and on some points, such as usage costs and timelines, the two gave different answers.
Staff had to search both documents, and nothing showed which answer was current.
What I Did
I went through both FAQs question by question. 49 questions appeared in both, and I combined each pair into one answer, giving a master FAQ of 118 questions.
Where the sources disagreed or a detail looked out of date, I did not pick an answer myself. I listed 5 decisions for the owner, each linked to the source behind it, and marked each one where it appears in the FAQ.
I formatted the master FAQ in the same style as the company's own team manual.
I set up an AI assistant in the company's CRM that answers only from the approved knowledge base. When it does not know, it says so and points to the support inbox instead of guessing.
I designed a branded page for the assistant that works without a login, built a QA tracker with a row for every question, and tested each one against the assistant. I also recorded a captioned demo video for the team.
Once the owner settled the 5 decisions, I updated the master FAQ and both original FAQ documents so all three give the same answer, added the company's published training lessons to the master FAQ, and loaded the final version into the assistant.
I rebuilt the page to match the company's main website, with the same header, colours, logo and mobile menu. The sample questions on the side can be clicked to send them straight into the chat. After launch I fixed a flicker where a placeholder picture showed for a moment before the company's icon on each reply.
What Went Into the Work
One master FAQ
118 questions from two documents, with the 49 duplicates merged into single answers.
A decisions list
5 conflicts or out-of-date details, each linked to its source for the owner to settle.
An AI assistant
Answers only from the approved knowledge base and sends anything else to support.
A no-login page
A branded page the team can open from any browser.
A QA tracker
A row for every question, recording the assistant's answer and the result.
A demo video
A captioned screen walkthrough showing the team how it works.
Matching source documents
Both original FAQs updated with the owner's decisions so they agree with the assistant.
Why It Matters
The assistant can only answer from what it is given. Because the documents were merged and the owner settled the conflicts first, it gives one answer to each question, the approved one.
The team gets an answer in seconds instead of searching two documents, and every answer was tested before launch.
The owner reviewed the assistant and the page and approved them, and the team now uses it from any browser. The two original FAQ documents match the master FAQ, so staff get the same answer whether they ask the assistant or read a document.
Tools and Skills Used
FAQ auditing, content merging, AI assistant setup, knowledge base design, landing page design, QA testing and demo video production.
Who This Is For
This case study may be useful to you if you are:
- ●A company whose support answers are spread across several documents.
- ●A team planning an AI assistant that wants the source content cleaned up first.
- ●A team looking for someone who tests every answer before an assistant goes live.
Need help with similar work?
Tell me where your support answers live today and who needs to use them. I audit the content and set up an assistant within the scope we agree.
Book a free call