Tech • AI • Robotics • Game
I Helped a Subscriber Build an App in 24 Hours
4/10TL;DR
A solo interior styling business in Manchester received a working custom CRM in under 24 hours after its requirements were narrowed to the smallest set of features needed for daily use.
KEY POINTS
Scope, not coding, defined the project
The business owner behind Willow Lane Interiors started with only a rough idea and no specification, wireframe or technical plan. The central challenge was not generating code quickly, but deciding which problems the first version had to solve immediately and which could wait. That approach turned a broad CRM concept into a practical one-day build.
The real problem was fragmented follow-up
Two lost jobs exposed the business risk: client follow-ups were spread across WhatsApp, a paper notebook and an outdated spreadsheet. A simple reminder tool would not have solved that, because it would have created a fourth place to check. The app was therefore designed around one operational question: who is waiting on her.
Version one was limited to four core functions
The first release focused on four essentials. It stored client records with name, phone, email, address, inquiry source, inquiry date and notes. It used a five-stage board, tracked quote values in pounds, and assigned a next follow-up date to each client, with overdue cases surfaced in a single list. Each record also kept a dated activity log so communication history no longer had to live inside chat threads.
Five popular features were deliberately cut
Invoicing and payments, automated email, a client portal, calendar sync and multi-user logins were all deferred. Each was dropped for a concrete reason: they required decisions or workflows the business had not yet defined, or they solved problems the owner did not currently have. The cuts helped ensure the build could be finished and adopted quickly.
The pipeline reflected the owner’s actual workflow
Rather than importing a generic sales template, the board was built around the business’s own five stages: new inquiry, consultation booked, quote sent, project live and completed. That meant the software fit existing habits instead of forcing the owner to adapt to a standard CRM model. The goal was to improve consistency without changing how the business already thought about its clients.
Design choices were intentionally non-corporate
The interface was specified as warm and calm rather than dashboard-like. The visual direction used an off-white background, soft near-black text and a deep fern green accent, closer to a stationery brand than typical business software. Just as important was the reject list: no blue or purple gradients, no dark mode, no glass effects, no oversized rounded buttons and no stock illustrations.
The first build included realistic sample data
The system was generated with 12 example clients spread across all five stages, using British names, Manchester-area addresses and quote values between £600 and £9,000. That allowed the owner to judge the workflow under realistic conditions rather than reacting to empty screens. The first version was shared quickly so real use could expose weak points.
User testing uncovered a critical omission
Once real clients were added, one practical flaw emerged: the app allowed a record to be created without a follow-up date. That meant a client could disappear from the overdue list entirely, recreating the very problem the app was meant to fix. The issue was not a software malfunction but a missing requirement, showing how real-world use reveals assumptions neither builder nor client had stated explicitly.
Refinements focused on actionability
Later changes made the overdue list more useful by surfacing phone numbers directly, so follow-ups could happen without opening each record. Column headers were changed to show total quote value rather than client count, because £8,000 in one stage matters more commercially than the number of names there. Overdue status was also made clearer on client cards, drag-and-drop movement was retained, and the sample clients were removed before handoff.
Branding and handoff helped make the app feel real
The default app address was renamed to WillowLaneCRM.base44.app, replacing a generic test link with the business name. The owner confirmed the system worked on mobile and said the functionality matched expectations. That final handoff turned the build from an experiment into a live operational tool the business could continue refining over time.
CONCLUSION
The project showed that rapid AI-assisted app creation depends less on coding speed than on disciplined scoping and close user feedback. By cutting nonessential features and building around an existing workflow, a solo business moved from an unformed idea to a usable custom CRM in less than a day.
Ask a question