Claude · Foundations

A company knowledge base is the difference between a smart guess and a right answer.

Ask a generic AI tool a question about your business and it will answer anyway, confidently, from nothing. A company knowledge base is what stops that: the actual services, pricing rules, policies, and terms your business runs on, written down once so the AI answers from what's true instead of what's plausible.

The problem

A blank AI tool doesn't know your business, and won't say so

The gap isn't the model. It's that nobody ever told it who you are.

Open a generic chat tool and ask it about your diagnostic fee policy, your service area, or which plan your best customers are usually on, and it will give you an answer. Not because it knows, but because generating a plausible-sounding answer is what it does when nothing stops it. That's the quiet failure mode of AI in a small business: not that it refuses, but that it guesses and doesn't flag the guess.

A knowledge base is the fix, and it's simpler than it sounds. It's not a wiki or a new piece of software. It's the specific, true facts about your business, written down in one place the AI actually reads before it answers anything.

Direct answer

What actually goes in a company knowledge base

Services and pricing logic

Not just a price list: the rules behind it. When a fee gets waived, what's included by default, how a quote changes with distance or scope.

Policies and boundaries

Cancellation terms, warranty rules, what staff are and aren't authorized to promise a customer without checking first.

Tone and terms your team uses

The actual words you use for your services and customers, so drafts sound like your business instead of a generic template.

Key documents and SOPs

The intake checklist, the service agreement template, the "how we handle this" procedures that currently live in one person's head.

What this looks like

The same question, with and without a knowledge base

An illustrative example, not a real business.

Without a knowledge base

Question: Do we credit the diagnostic fee toward a repair?
Answer: Many businesses in this industry do offer that. It's common to apply the fee if the customer proceeds with the repair.

With a knowledge base

Question: Do we credit the diagnostic fee toward a repair?
Answer: Yes, toward either a repair or a replacement, as long as the customer books within 30 days of the visit. Exact wording is in the pricing policy document.

The first answer is confident and wrong for your business. It's a fair guess about the industry in general, offered as if it were fact. The second is boring by comparison, and correct, because the workspace was told the actual rule instead of being left to infer one.

Where it comes from

You already have most of it. It's just not in one place.

Almost nobody starts a knowledge base from a blank page. The material already exists: a pricing sheet, a service agreement template, the answers an experienced office manager gives from memory a dozen times a week. Building the knowledge base is mostly a matter of collecting it, resolving the places where two people would answer differently, and writing down the judgment calls that never got written down.

A knowledge base is never finished. It gets updated as pricing changes or a new policy gets set, the same way you'd update a price sheet. What matters is that it's the one place the AI, and your new hires, can both check.

Where it fits

This is session 1 of the AI Concierge

The knowledge base isn't a separate project. It's the foundation everything else in a Claude workspace gets built on.

Before any custom skill gets built, the first working session loads your company's actual services, policies, pricing rules, and terms into your workspace. Every skill built after that, an estimate follow-up, a document summary, a customer reply, draws from the same knowledge base, which is why the drafts sound like your business instead of a generic one.

Questions

FAQ

How is this different from a website FAQ page?

A website FAQ is written for customers and kept short. A knowledge base is written for the AI and your team, and it includes the internal judgment calls a customer-facing FAQ never would.

How long does it take to build one?

Through the AI Concierge, the first version gets built in a single working session, from an intake questionnaire and your existing documents. It's refined from there as your business changes.

What if two staff members would answer a question differently?

That's usually the most useful thing building a knowledge base turns up. Resolving it once, in writing, is worth more than the knowledge base itself.

Next step

Let's see what belongs in yours

A short call to look at what your team currently answers from memory, and what a knowledge base built around your business would look like.

Start with a conversation

Tell us where the week goes. We'll show you what to fix first.

A 30-minute call, no pitch deck. You describe the work that eats your time, and we tell you plainly whether AI is the right fix, and if so, where to start.