October 2026: our starting point
Prepared on 4 October 2026. Correction added 18 October: the old provider amount below was GBP, not USD; it is retained as historical context. The maintained scorecard now uses the later reconciled billing export. This is the first draft of the proposed Open Company update. It covers the operating-system work and the dated evidence available to it, not a live monthly close.
Conceptual illustration. It does not show the Pineapple interface or a measured result.
We want to find out whether AI can help Pineapple serve customers and run with less avoidable work from Peter. We already use AI for investigation, reports, and preparing changes. The missing piece was a clear account of the rules, evidence, and results.
What this work produced
A connected handbook explaining our customer focus, decisions, metrics, AI boundaries, marketing process, and privacy rules. It also includes a starter workflow and templates other business owners can use.
The guide builds on Peter Assistant's existing records. It does not replace our business plan or create another experiment database. The local export process includes source-revision checks so a changed internal decision cannot quietly leave an old public draft looking current.
What the business evidence says
The September review records provisional ProfitWell MRR of £1,524.88 GBP and 126 active customers. That is a provider view, not a reconciled company total. Billing coverage and late cohort windows still limit the conclusions we can draw. The scorecard explains those limits beside the numbers.
This matters for AI: an agent cannot make a reliable growth decision by turning an incomplete provider count into a company-wide fact.
What is still unproved
We have not measured a company-wide autonomous completion rate, time saved, or cost per verified AI result. We have not proved that this guide improves customer understanding or paid growth. No customer outcome should be attributed to this documentation work yet.
The public guide and any new live execution permissions remain separate release decisions. Drafts and local checks do not establish that content is saved in the production CMS or visible to customers.
Our next learning step
After release approval, ask three suitable business owners to find who Pineapple is for, explain which actions require Peter, and complete the starter workflow. Record where they hesitate. This small review can expose unclear writing; it cannot estimate a population-wide conversion rate.
The next company update should include a verified business snapshot, one real decision, a failed or inconclusive attempt if there was one, the measured human work, and what changes next. If a value is missing, say why.
Start with the operating loop, or try the workflow in your own business.
Updated 19 October 2026.
