AI Knowledge Layer.

We help you consolidate your operating logic into a platform that your teams and agents can use.

This is for you if...

Your AI can access your data. It still has no idea how your business actually works.

Your agents make mistakes that any experienced employee on your team would never make.

If your best person left tomorrow, some of what they know would leave with them.

You have tried to automate a process and the AI kept getting the edge cases wrong.

Your team and your AI are operating from different playbooks, and neither knows it.

What you get from a knowledge layer engagement.

Consolidated SOPs

The rules your team follows without writing down, codified through structured sessions.

How does check-in actually work?
3:42

“If they land after nine you text the lockbox code, but never before the cleaner's signed off.”

Head of Ops
Guest check-in
Cleaner has signed off the unit
Arriving after 21:00? Text the lockbox code
Otherwise, meet them at the door

We help you run sessions to record the standard path through each process alongside the exceptions people actually make. This will help in creating SOPs for the key processes, reducing dependency on key team members.

Human and machine readable knowledge base

A single knowledge base for your entire business operations.

processguest_check_in
steps7
ownerOps
formatSOP
last_verified30 days ago
retrieval_indexkb-ops-001

We build a shared platform for SOPs and process definitions, that powers a human readable version for your teams, and a structured version for (AI) systems. Changing a rule updates both at once to ensure consistency across applications.

AI tool connections

Your existing AI tools wired into the knowledge base so they read the rules instead of guessing.

AI tools
Copilot
ChatGPT
Internal bot
Outputs
Precise answers
Correct actions
Consistent docs
Knowledge layer

We work with you to connect the assistants, agents or automations you already run. Each connection is limited to the part of the knowledge base that tool actually needs.

Maintenance guide

What to do when a process changes, so the knowledge base changes with it.

ActionOwnerFrequencyNext
Update process docs
OpsMonthly
Mar
Re-verify definitions
RevenueQuarterly
Q2
AI accuracy check
DataMonthly
Mar
Add new processes
Team leadsAs needed
-

We draft the process to update the knowledge base when a rule changes, and how that change reaches the connected tools. It also handles the reverse case, where an AI output starts going wrong because the documented rule went stale.

Case study.

Unified metrics, definitions and SOPs

Short-term rental operator, 100+ properties

99.9%data accuracy (up from 91%)

A short-term rental operator with 100+ properties had connected their stack: PMS, chatbot, WhatsApp, Google Sheets, all in one place. But the business logic underneath had never been written down. Metric definitions didn't match across systems. Teams weren't working from the same numbers, and neither were the first few AI workflows. We built a knowledge layer to solve this, with clear ownership of every definition, shared context between human teams and AI agents, and a direct API connection to Claude for querying occupancy, revenue, and guest data.

What changed

  • Business logic codified and shared across human and AI teams
  • Direct data querying via Claude for occupancy, revenue, and guest insights
  • Claude workflows enabled for pricing analysis, guest communication, and reporting

How we help you.

Start with a free call. From there we either run a full value iteration on your stack and hand you the plan, or we stay on as your fractional CTO for data and AI.

  • Checkpoint

    Free, 30 minutes

    Bring one thing that's stuck. We tell you what's wrong with it and what to do first.

    Book a checkpoint

  • Value Iteration

    Scoped engagement, one-off

    We analyse your whole stack and build a pragmatic plan: what's broken, what to fix first, and who should own it. Your team or any vendor you hire can run it.

  • In the Loop

    Ongoing, fractional

    We work with you a few hours a week on what moves the strategy: prioritisation, vendor calls, architecture decisions and hiring.

Questions we get asked.

The process logic, decision rules, and institutional knowledge that currently live only in people's heads: how your team handles edge cases, what good output looks like, which exceptions to the standard process exist and why. Structured so AI systems can read and follow it.

SOPs are written for humans. They assume context, tolerate ambiguity, and get ignored. An AI knowledge layer is written for machines: precise, structured, with explicit rules for every case. The output is process definitions a machine can read.

Traditional documentation fails because nothing depends on it staying current. A knowledge layer feeds directly into your AI systems, so when a process changes and the documentation does not, the AI starts producing wrong outputs. That is your signal to update.

Through structured extraction sessions: part interview, part process walkthrough. We observe the work, ask the questions that surface unspoken rules, and convert the answers into structured definitions.

Repetitive, judgment-heavy processes that currently depend on one or two people who just know how it works. Fulfilment logic, escalation handling, content review, pricing exceptions. Anywhere a key-person departure would break the operation.

Let's get started.

Book a checkpoint