Claude, explained
Discovery call · 15 Jun 2026 · Simon · Jigar · Andrew

Building an intelligence layer over everything you already have.

The call distilled to the decisions that matter — what Simon's team wants to build, where the data lives, the approach Andrew would take, and how an engagement would actually start. Read top to bottom.

01 What you want to build 02 Where the knowledge lives 03 Tidy first, no database 04 Every folder is an agent 05 Cowork vs Claude Code 06 The agents you want 07 The finance agent 08 Who runs it 09 How we start
01

What you actually want to build

Right now the team are heavy users of ChatGPT and Claude — projects spun up, documents loaded in, answers bounced between tools. Useful, but as Simon's boss put it, that's twenty chatbots doing twenty jobs.

The goal is an operating system above them: an intelligence layer one or two people can ask anything across the whole business — plus a handful of discrete agents for the thorny functions. Realistically three or four: clients & new business, finance & operations, and so on.

One layer to ask anything, a few agents underneath
Intelligence layerask anything across the business · 1–2 people
Clients & new business Finance & ops M&A research + more
your data — Drive · Xero · Streamtime · email
Not one mega-bot — a thin layer on top, with a few focused experts beneath it.
02

Where the knowledge actually lives

A 140-person marketing group across London, the US and expanding into Asia — and the knowledge is scattered. About 90% sits in Google Drive, in ordinary drives. The rest is the usual legacy spread: some on Dropbox and a legacy Microsoft system, mostly used by finance.

Accounting runs on Xero (seven companies); projects live in Streamtime — which doesn't yet talk to accounting, so figures get exported and re-keyed by hand.
Email is mostly Gmail, but some people — Simon included — are still on Outlook. Slack is for chat, not documents.
Many sources, one place to ask
Google Drive 90% Dropbox Microsoft · Xero Streamtime Gmail · Outlook
one place
to ask
03

Biggest bang for the buck: tidy the docs, skip the database

The cheapest, highest-leverage move isn't new infrastructure. It's taking the existing folders and documents and making small tweaks — a couple of map files, cleaner naming conventions — so an AI agent can explore them the way a new hire would.

Talk to enough people and you'll hear two approaches. A database has real benefits but a lot of setup and cost. The lighter path: structure what you already have, and keep the database in the back pocket until something proves you need it.

Start here Structure the folders

A few files and naming rules so Claude can walk the drive: this folder is clients, each has meeting notes, each note is named by its date. Low effort, fast to value.

Back pocket Build a database

Powerful, but heavy on setup and ongoing cost. Reach for it only if the simpler approach hits a wall — not as the default first step.

In an ideal case, it's just a couple of files and some renaming — not crazy amounts of effort.
04

Every folder becomes an agent

The architecture is a discovery route. At the top sits a simple map — there's a clients folder, a finance folder, an operations folder. When you ask something, Claude reads the map, then goes folder by folder, loading a specific file in each so it understands the context.

Ask to change a statement of work for a client and it knows to open that client's folder, load the context, and become the expert on them. Bite-sized steps are the whole trick — they keep the context small so Claude never gets overloaded.

Read the map, walk the folders, load the context
Clients Finance Operations
CLAUDE.md becomes the
Simon expert
For a whole-company layer, work in small bite-sized pieces so you never overload the model.
05

This is Claude Code, not Cowork

Cowork is friendlier, but it runs locally on one machine, which makes it hard to share context across a team. For a layer the whole company leans on, Claude Code is the fit — it's what walks the folders on that discovery route.

For this Claude Code

Runs the folder-by-folder discovery, manages context in small pieces, and is the better base for something a team relies on.

Limited here Cowork & n8n

Cowork is local and harder to share in a team setting. n8n likely plays only a small part in this build.

06

The agents you'd actually want

Beneath the top layer sit the working agents — each scoped to one function, with access only to the folders it needs. A few that came up directly on the call:

Clients & sales
The person writing a SOW can ask it to check every past SOW and suggest how to improve this one.
Finance & ops
Compares year-end statements against Xero, and runs the AP / AR invoice flow across the seven companies.
M&A research
Pulls comparables for a target — easy from Companies House in the UK, harder in France, where it has to find the filing body itself.
New-brief intake
Captures briefs arriving by email — possibly via one dedicated address — so nobody hunts through inboxes for them.
07

The finance agent, concretely

Jigar's clearest example: a finance agent watching the accounts-payable and accounts-receivable inboxes. Set up roughly twenty rules; once an incoming invoice is authenticated against them, the agent flags it, logs the activity, picks up the invoice and posts it into Xero — to the right one of seven companies.

Invoice in, rules checked, posted automatically
AP / AR inboxan invoice arrives
~20 rulesauthenticated & flagged
Posted to Xeroright company · 7 of them
The exact rules get defined once the process flow is mapped with finance — the shape is clear today.
08

Who runs it — and it's both

Two ways to keep this alive: teach AI fluency across the org so people set up their own agents, or have Andrew act as the part-time AI person. The honest answer is a mix — he'd be on for the journey, six to nine months, while people get up to speed. Not a train-and-leave.

You'd want one power user / project manager as the main person around it, plus time with the people whose processes are being automated — a lot of it lives in their heads.
An IT manager handles access. Andrew develops; once the process flow is clear, he asks IT and ops the right questions about permissions and where each agent should look.
09

How an engagement would start

What makes or breaks the timeline is the amount of data and how well structured it is — getting things into shape can be most of the work. The plan: a short paid discovery first, and if it isn't viable, the money comes back.

Discovery~1–2 days, spread out
Map the process flow with Simon, finance, ops and IT. If it's not going to work, Andrew refunds it and says so.
Focused build~3 months
Structure the drive, stand up the intelligence layer and the first agents. Estimate firms up after discovery — could be two months, could be more.
Ongoinglight retainer
A couple of hours on an ongoing basis — new needs surface and things need maintenance as the business uses it.
Distilled from the 15 Jun 2026 discovery call · Simon · Jigar · Andrew Back to the explainers