OneLife: Engineering AI-powered Personal Growth Platform
How an AI self-development product engaged a full team consisting of a Solution Architect, a Business Analyst, a Project Manager, a UX Designer, 2 Full-Stack Engineers, a Mobile Engineer, and a QA Engineer.
Challenge
OneLife is a personal growth platform built on an unusual idea: the user chooses the books that have shaped them, and the system learns to answer in their voice. Underneath that sits more machinery than the description suggests. Text has to be ingested and embedded into a private knowledge base per user, spoken reflections have to be transcribed and read for emotion and significance, behaviour has to be clustered into a growth archetype, and the whole thing has to come back as something that sounds like guidance rather than output. A product like this can be specified a hundred different ways, and the gap between the best and worst of those specifications is a year of engineering. OneLife had the idea, and no team of any kind.
So the first four requests were not for engineers. A Solution Architect to decide the architecture and the AI pipeline. A Business Analyst to turn an idea into a traceable backlog. A UX designer, six years in Figma with design systems and interactive prototypes, for an interface built on symbolic metaphor and motion rather than on screens. A Project Manager with twenty years of delivery, used to turning business requirements into epics and user stories and running agile at scale. Four requests to answer what was being built, in what order, and why, before anyone wrote code that would have to be thrown away.
The build team of another four followed, specified against what the first four had produced, while keeping a UX designer and a project manager. Two full-stack engineers on Node.js, NestJS, React and TypeScript, one of them with microservice and account-system work behind them. A mobile engineer with five years of engine-level development, performance optimisation and legacy refactoring. A QA engineer with five years owning the full testing lifecycle across functional, UI, regression, API and localisation testing, on Jira and Postman. The first engagement was confirmed 48 hours after the request opened.
Outcome
OneLife organised the work in the following way: architecture, requirements, design and delivery management first, as a team rather than as a document, and the engineers afterwards, briefed against a phased plan that ran from proof of concept through MVP to scale.
- Eight roles across two stages, with the first match confirmed in 48 hours.
- Definition and delivery both covered: architecture, requirements, design and delivery management, then the engineering and testing to build against them.
- A complete product team where there had been none, with the definition roles ending when their work did.
- Nine months on, the full team is working through a phased plan from proof of concept to scale.
I have only positive feedback. Communication was smooth, we easily found a common language, and they clearly understood the vision and direction of the project.

A Strategic, Multi-Phase Hiring Approach
Phase: Request
Two planned stages rather than one: four requests to define the product, covering architecture, requirements, design and delivery management, and then four more for the engineers and the QA who would build it, specified against what the first wave produced.
Phase: Matching
The architect was screened on architecture owned through to production rather than handed over at the diagram, the analyst that defined working systems in the wellbeing space, and the build team on production work in the stack the architecture called for, with 8 of 32 shortlisted going through to interview.
Phase: Interviews
The founder ran the interviews and personally made every decision to hire, and the first engagement was confirmed 48 hours after the request opened.
Phase: Engagement
Engagement opened with a mix of full-time and part-time monthly workloads, with the definition roles sized to the work in front of them, daily hour commitments in the tracker and monthly reports with detailed breakdown of hour commitments for approval.
Phase: Support
Engagements have run for nine months, with workloads and hour caps changing between phases and every change being approved by the founder.
