A working note on building alira.london in public. The automation, the mistakes, and the parts I have not talked about until now. --- There is a version of this article that is clean and professional. Lessons learned. Key takeaways. A tidy list of what worked and what did not. This is not that version. I have been building alira.london for the past several months. The honest account includes sleepless nights, lost weekends, and a period where I actively did not want to be around other people. Not because I was struggling in a way that worried me. Because I was so deep inside the build that human conversation felt slow. Unstructured. In the way. The only thing that felt productive was opening a new chat with Claude. --- What alira.london Actually Is The original idea was simple: most small businesses do not have access to decent strategic thinking. A consultant costs money most of them do not have. So I built a diagnostic at alira.london. Six questions that force a real answer, and a personalised business plan built from whatever someone actually told us. That part worked. People complete it. They read the plan. A few of them come back. The problem was everything around it. The tools, the automation, the content pipeline, the payment flow. Each one looked fine in isolation and was quietly broken in practice. --- The Part I Did Not Talk About At some point in the last few months, I stopped making plans for weekends. Not consciously. They just stopped happening. Saturday became the same as Tuesday except there were fewer messages to distract me. I would open the laptop at eight in the morning and look up at seven in the evening and feel nothing except the urge to keep going. I stopped texting people back quickly. Not because I did not care. Because I was mid-thought and coming out of it felt like a loss. Somewhere in that period, my relationship with AI shifted. It stopped being a tool I used occasionally and became something closer to a daily anchor. Every morning, before anything else: open Claude, pick up where I left off. It was structured in the way that prayer is structured. Not spiritual. Ritualistic. A way of beginning the day with something that did not need small talk and would not judge the fact that I was working on the same problem for the fourteenth morning in a row. I am not saying that is healthy. I am saying it is what happened. --- The Automation Experiments I have been trialling automation across three areas: content, outreach, and the product itself. Content was where I started. The idea was to generate business strategy articles automatically, post them to LinkedIn and Google Business Profile, and build organic search traffic without manually writing three articles a week. That part mostly works. What did not work was the tone. The early posts were written as a curious observer. Someone noticing things about business from the outside. Safe, generic, forgettable. The kind of content that earns no engagement because it asks nothing of the reader and proves nothing about the author. I fixed that recently. The posts now reference real decisions I have made using alira.london's own tools, and rotate through a build-in-public format where I describe what a SWOT or 5 Whys run on the business itself actually produced. I should have done it from the start. The reason I did not is that the honest version requires you to say something real. And for several months I was not ready to do that publicly. The contact form was sending nothing. It had a loading state, a success message, the right fields. No enquiry reached me for weeks. I found out during a full audit of the site. The fix took ten minutes. The cost was the enquiries I never received. The payment screens on the tool paywall were the same. Nine separate screens. Each one had a button that looked functional. None of them linked anywhere. A user who wanted to pay could not. I had built a paywall that nobody could get through, and I had been too absorbed in the next feature to notice. --- What the 5 Whys Actually Told Me I ran the 5 Whys on the checkout drop-off rate after I found both of those issues. Why were users not completing the purchase? The button did not work. Why did the button not work? It was never wired to the payment flow. Why was it never wired? I was building features sequentially and had not reached the payment integration. Why had I not prioritised the payment integration first? I was solving product problems before commercial problems. Why was I solving product problems first? Because product problems feel more concrete and less exposing than asking someone to pay. The fourth why is the one that mattered. I was building things I found interesting ahead of things that made the business work. That is what happens when you are alone with a build for too long. There is nobody to tell you that you are going in circles. --- What I Have Learned Three things about automation. One thing about isolation. Automation amplifies whatever is already there. If your contact flow is broken, automating around it makes it worse. Fix the thing before you automate it. The audience follows the content, not the platform. I built a LinkedIn posting pipeline and pointed it at the wrong people. Most of my connections are NHS professionals from earlier work. Business strategy content does not land with NHS teams. The content was not the problem. The network was. You cannot automate your way out of an audience mismatch. The PDSA cycle is genuinely useful when you are honest in the study step. Plan a change, make it, study what the data says, adjust. I ran it on the homepage, the contact form, the LinkedIn post format, and the paywall screens over the past month. Most of the study step produced uncomfortable answers. That is how you know it is working. Building alone with AI as your main thinking partner is efficient and corrosive in equal measure. The efficiency is real. You can move fast, work through problems at two in the morning, operate at a pace that is not possible with a traditional team structure. The corrosion is also real. You stop reality-checking. You lose perspective on what is actually broken because nothing external is giving you friction. The broken contact form, the dead payment buttons. These are things another person would have caught in an afternoon. I missed them for weeks because I was deep inside features nobody had asked for. --- Where This Is Going alira.london is a consultancy that also has a tools platform. The consultancy side is where the real work happens. The platform is proof of how we think. The next phase is simple: stop building features, start improving conversion. The diagnostic works. The plan output is good. The tools are functional. What I need now is for people who arrive to actually move through it. The content is changing. The LinkedIn posts are changing. This article is part of that change. The honest version is always better. It just takes longer to write. --- The diagnostic is free and takes about ten minutes. alira.london/create-plan