À la carte · Knowledge System

Everything your team knows that the AI keeps forgetting — written down once, in files you own.

Every AI tool you try starts from zero. Your prices, your process, your customers, the things you never promise — you explain them again, and the next tool forgets them too. A Knowledge System writes that knowledge down once, as plain files in an open format any AI tool can read. An AI keeps it current by proposing changes. A person you name approves them.

Request a Knowledge Audit See What's Included

The problem isn't the AI. It's that nothing is written down where the AI can read it.

Three things we hear in almost every small business that has started using AI tools.

01

Every tool starts from zero

The chatbot, the email assistant, the new hire's AI helper — each one is told the same things by hand, gets some of them wrong, and is told again.

02

It lives in two people's heads

What you'll discount and what you won't. Which customer gets a call, not an email. Why you stopped doing it the old way. None of it is anywhere an AI — or a new hire — can find it.

03

It's rebuilt for every new tool

Knowledge typed into one vendor's settings stays with that vendor. Switch tools and you start the setup again.

How it works: one fact, one file, a source and a signature

Your knowledge becomes a folder of short, plain-text files — one per topic — in the Open Knowledge Format, an open-source specification published by Google Cloud. Any AI tool can read them. So can you, in any text editor. Each file says where its facts came from, when it last changed, and who confirmed it.

policies/refunds.md
---
type: Policy 1
title: Refunds
description: When we refund, how fast, and who decides.
sources: 2
  - resource: "meeting:2026-08-14"
    title: Owner walkthrough
verified: { by: "human:owner", at: "2026-08-20" } 3
stale_after: "2027-02-20" 4
---
# Refunds 5

Full refund within 30 days, no questions.
After 30 days: the owner decides, case by case.
**Never** promise a refund by email before
the owner has seen the order.
An illustrative file, not a real client's.
1

One topic per file

This file covers one thing — refunds. type: Policy says what kind of topic it is, and the description sums it up in a line. Pricing, a product or a customer type would each get a file of their own.

2

Every fact has a source

sources shows where it came from: the owner walkthrough on 14 August. You can always check a fact against where it was said.

3

A person signs it off

verified records that the owner confirmed it on 20 August. Until someone you name signs off, a fact is labelled unconfirmed.

4

It knows when it's out of date

stale_after flags the policy for a re-check on 20 February. Prices, certifications and contacts get the same date.

5

The rule, in plain words

Below the header is the policy itself — including what never happens. That’s the line an AI tool reads before it answers a customer.

Propose, then approve

The AI does the upkeep. You decide what's true.

An AI agent reads what's new — documents, meeting notes, anything you drop in a folder — and drafts the changes. It never confirms its own work.

Talk to Us About Fit
  • New or changed facts arrive as drafts, with their source, for a named person to approve.
  • Nothing is deleted. A replaced fact is kept, marked as no longer current, and linked to what replaced it.
  • A weekly check reports what's out of date, what contradicts itself, and what has no source.
  • Payment details, credentials and anything you mark restricted stay out of it.
  • Medical, legal, tax and financial matters are recorded as your professional said them — the AI never advises on them.

Three steps, each one scoped before it starts

Start with the audit. Every step is scoped and quoted for your business before any work begins.

STEP 1

Knowledge Audit

We talk to the people who hold the knowledge and find where the rest of it lives. You get a list of every topic worth writing down, the proposed structure, the gaps, and the first three files written as samples.

STEP 2

Bundle Build

We write the files, link them, and set up the AI agent that keeps them current — connected to the sources you approve. It's tested on three real tasks you choose before it's handed over.

STEP 3

Maintenance Retainer

Optional and monthly. We review the weekly checks with you, add topics as the business changes, and keep the history of what changed and why.

The ROI Compass

Every recommendation moves one of three numbers

The ROI Compass is how Simplex Via ranks every recommendation, in any engagement. The audit asks what a wrong or re-discovered fact costs you today, in your own numbers — and the build is judged against that, not against an industry average.

Effectiveness

It Makes You Money

Quotes, follow-ups and answers that use your real prices and rules, the first time — instead of an AI guessing and someone fixing it later.

Efficiency

It Gives You Hours Back

The same things stop being explained twice — to each new tool, each new hire, each time someone forgets.

Quality

It Makes Customers Happier

Every tool and every person gives the same answer, because they're all reading the same confirmed files.

Is this for you?

A good fit if

  • You or your team already use AI tools, and keep re-explaining the same things to them
  • Important knowledge lives with one or two people
  • You've switched tools before, or expect to
  • Someone will spend a few minutes a week confirming what the AI drafts

Probably not yet if

  • You don't use AI tools yet — start with the free AI Value Self-Assessment
  • You want a chatbot built, not the knowledge behind it
  • Nobody can own approving changes
  • You need every historical document migrated — that's a different, larger project

What you walk away with

Yours from the first file. Plain text, an open format, in a folder or repository you control.

The knowledge filesYour business, one topic per file, sourced, dated and confirmed.
The AI agent's instructionsHow it reads the files, drafts changes, and runs the weekly check.
A one-page handoverHow your team adds to it and approves changes, written for people who won't read a manual.

Questions

Is this a chatbot?

No. It's what a chatbot, an email assistant or any other AI tool reads from. You can use it with the AI tools you already have.

What is the Open Knowledge Format?

An open-source specification, published by Google Cloud under the Apache 2.0 licence, for writing knowledge as a folder of plain markdown files with a small header on each. Because it's open, your files aren't tied to us or to any vendor.

Who owns it?

You do, from the first file. It lives in a folder or repository you control. If you stop working with us, it keeps working.

Does the AI change things on its own?

It drafts changes; it never confirms them. A person you name approves each one before it's treated as true.

Which AI tools does it work with?

Any tool that can read files. We set it up with the tools you use; the files themselves don't depend on any one of them.

What does it cost?

Each step is scoped to your business and quoted before work starts. The audit comes first, and tells us both whether the build is worth it.

Find out what your business knows that your AI doesn't

Pick a time and we'll talk through where your knowledge lives today and whether an audit is worth it.

Request a Knowledge Audit

Please don't send confidential customer information when scheduling. We'll agree what's in scope before asking for access to anything.