Latest / Solopreneur Sessions with Fexingo: One-Person Businesses, Freelancing, and Independent Work / How One Solopreneur Uses AI to Cut Client Onboarding Time by 80 Percent
Transcript
- Lucas: Mira Chen runs a solo brand strategy practice in Austin. She charges five figures per engagement, and for years her bottleneck was the onboarding phase — the first three weeks of every client relationship. Luna: Three weeks? That's a lot of unpaid prep time before the real work even starts. Lucas: Exactly. She'd spend those weeks doing deep-dive interviews, auditing existing materials, building a brand personality framework — all manual, all one-on-one. And because she's a solo operator, she could only take on two or three clients per quarter. Luna: So the constraint wasn't the quality of her work — it was the intake process. Lucas: Right. And about six months ago, she decided to experiment with a custom AI agent to handle the first 80 percent of onboarding. Not the relationship part — the data gathering, the analysis, the pattern recognition. She built what's essentially a guided questionnaire that adapts to each client's answers, powered by a large language model trained on her past 40 projects. Luna: Wait — she trained the model on her own past client work? That's proprietary data. How did she do that without handing everything over to OpenAI? Lucas: She used a combination of a custom GPT inside ChatGPT plus a secure vector database on her own server. The model never stores the raw data — it just retrieves relevant examples from her past work when a client gives a certain answer. So it's essentially a RAG system — retrieval-augmented generation — but she doesn't call it that; she just calls it 'the bot.' Luna: And clients actually engage with a bot for the first week of a high-ticket engagement? Lucas: That was her biggest fear too. But she frames it differently. Instead of saying 'talk to my AI assistant,' she sends a personalized welcome note that says 'I've set up a smart workbook that will help me understand your brand faster. It asks questions you'd normally get in a two-hour call, but you can answer on your own time, with your team, and it'll adapt based on your responses.' Clients love it. They feel it's more efficient. Luna: So it's essentially asynchronous depth — but with a human touch at the start and end. Lucas: Exactly. The bot collects initial brand history, target audience profiles, competitor references, visual preferences — then it spits out a preliminary brand personality framework and a list of gaps or contradictions in the client's answers. Mira reviews that output in about an hour, then jumps on a single two-hour call to validate and refine. Luna: So the three weeks of back and forth turns into one call and a little bot homework. Lucas: Right. She now gets from signed contract to approved brand strategy in four business days. That's an 80 percent reduction in onboarding time. And here's the thing — her clients actually report feeling more heard, not less. Because the bot captures every single answer, no notes get lost, no nuance dropped. Luna: That's the counterintuitive part. Automation can actually improve the human experience if it's designed to capture detail rather than shortcut it. Lucas: And the business impact is real. She went from three clients per quarter to five, with no increase in her working hours. Revenue jumped about 60 percent. But she's careful not to call it scaling — she still only serves five clients per quarter because she wants to maintain the personal creative work. Luna: So the AI didn't replace her — it made her more available for the parts where she actually adds unique value. Lucas: That's the whole thesis. And I think it's a model that many solopreneurs can adapt. If you have a repeatable information-gathering phase in your service — interviews, audits, data collection — you can probably offload 70 to 80 percent of it to a custom agent without sacrificing quality. Luna: But there's a risk. If you train the bot on bad past projects or biased data, you're just automating your worst habits faster. Lucas: Absolutely. Mira spent two weeks cleaning her own past deliverables — removing outdated frameworks, correcting language that was too jargon-heavy. She also built in a confidence threshold: if the bot's output doesn't meet a certain score, it flags the case for human review. So the system knows when to punt back to her. Luna: That's smart. It's essentially a fail-safe for edge cases. Lucas: Yeah. And the cost is minimal. She pays about $60 per month for the custom GPT plus the vector database hosting. The time savings alone is worth that many times over. Luna: It reminds me of something I've been thinking about with our own show. A lot of listeners tell us these episodes give them practical ideas they can try right away — and that kind of value is hard to put a price on. Lucas: That's actually something we hear a lot too. And it's why listener support matters so much. A couple of dollars a month is genuinely what keeps these going — buy me a coffee dot com slash fexingo, if you've gotten something out of them. Luna: Totally. It's people like Mira who inspire us to keep digging into these stories. And the support makes it possible to keep the show ad-free. Lucas: Alright, back to Mira. The other thing she did that I think is worth talking about is how she priced this faster onboarding. She actually raised her rates. Luna: Because she could deliver faster results? Lucas: Partly. But also because the bot gave her a data-driven edge. She now shows potential clients a sample output from the bot — a mini brand audit based on their website and social media — before they even sign. That's a teaser that used to take her a full day to produce. Now it takes ten minutes. Luna: So the same tool that speeds up onboarding also becomes a sales tool. Lucas: Exactly. Her close rate went up by about 25 percent after she started sending those mini audits. Because prospects see the depth of analysis before they commit. Luna: That's a great loop. Faster onboarding, better sales, higher rates — and the client still gets a human strategist at the end. Lucas: Right. And I think the lesson is not 'build an AI bot.' It's 'find the repetitive intellectual work you do and figure out if a machine can do the first pass.' Luna: For solopreneurs, the bottleneck is often the thinking that feels bespoke but actually follows patterns. And that's exactly where AI shines. Lucas: Mira's now experimenting with a second bot for the middle of projects — ongoing brand maintenance and content suggestions. But she's doing it slowly. She wants to keep the human element in the creative decisions. Luna: Which is the right instinct. The goal isn't to automate away the soul of the business — it's to automate away the drudgery so you have more energy for the soul. Lucas: Well said. That's a good place to leave it for today.