Projects
End clients are described by sector: their names and their figures are theirs, and they stay that way. Anything public carries a link. Below the cases, the process I work any project in.
Odoo with OCA modules on infrastructure as code, with continuous integration, observability and disaster recovery that was tested rather than assumed. The platform can be rebuilt from its repository, which is what makes a mid-sized company safe to upgrade.
For a consumer-goods multinational: target operating model, gap analysis and integration decisions written down before anything was configured, modelled in C4, BPMN 2.0 and ArchiMate. A React application let the people who run each process read their own flow instead of taking the architect's word for it.
Design and proposal for a CRM tied into WhatsApp, where customers in this part of the world actually write, with incoming opportunities classified automatically. Delivered with a business case and an estimated return.
A public prototype: from the transcript of a discovery meeting it produces user stories with acceptance criteria. The first piece of a portfolio of agents for the software lifecycle.
See it on GitHubMy own Odoo Community with a custom theme, self-hosted on a Mac mini behind a Cloudflare Tunnel: website, customer portal, CRM, quotations, invoices and the digital product in a single system, with encrypted backups kept off the premises and zero recurring cost beyond the domain. It is the same argument the site makes, run on myself first.
Before proposing anything I survey what is already there: processes, systems, data and who decides. The output is a gap map the client recognises as their own. Process mapping
Target model, integrations and decisions recorded before a single setting is configured. What is not written down has not been decided. Architecture map
Configuration, prototype, user acceptance and go live, with clear closing criteria at the end of every phase.
Living documentation, training module by module and adoption measured: the capability ends up in the team, not in the consultant.
Herbert Simon called it satisficing: set the level a solution has to meet, then look for the alternative that meets it. There is no need to maximise effort, time and cost when the solution has to fit the resources and the deadline the client actually has.
That is how I design. An explicit aspiration level, written down, evidence to know when it is met, and the discipline to stop. The maximising system arrives late and sometimes never arrives at all: it runs out of budget, out of time, and ends up being more than the organisation can carry.