How I Use AI to Run 3 Businesses

Three Businesses, One Person, Not Enough Hours

I run three businesses. A management consulting practice that serves mid-size companies. An e-commerce store selling specialty office supplies. And a content business that produces training materials for corporate clients. None of these are massive. The consulting practice has two associates and a part-time admin. The e-commerce store is me, a warehouse partner, and a freelance designer. The content business is mostly me and a contract editor. Before AI, I was working 60-hour weeks and still dropping balls. Emails piled up. Client reports were always late. Product descriptions were generic. Training materials took weeks instead of days. Now I work about 45 hours a week. The businesses generate more revenue than they did at 60 hours. Here's exactly what changed.

6:30 AM — The Morning Triage

My day starts with what I call the morning triage. I open my inbox — usually 40 to 60 emails across all three businesses overnight — and I use AI to process them in batches. For the consulting practice, I paste client emails into Claude and ask it to draft responses. Not final responses — drafts. About 70% of the time, the draft needs only minor tweaks to match my tone. The other 30% need real thought, but AI still saves me from staring at a blank reply box. For the e-commerce store, I have a standard prompt that sorts customer service emails into categories: shipping issues, returns, product questions, and complaints. AI drafts replies for the first three categories. Complaints always get a personal response from me. This morning triage used to take 90 minutes. Now it takes about 35. That is not a guess — I tracked it for two weeks when I started.

The Consulting Practice — Research and Reports

The biggest AI win in my consulting practice is client research. When I'm preparing for a new engagement, I need to understand the client's industry, competitors, recent news, and market trends. Before AI, this was a full day of reading annual reports, industry publications, and news articles. Now I still read those sources — AI did not replace the reading — but I use AI to summarize long documents, extract key data points, and identify patterns I might miss. A typical prep session: I'll feed an annual report into Claude and ask for a summary focused on operational challenges. Then I'll ask it to compare those challenges against industry benchmarks I already know. Then I'll ask it to draft discussion questions for my first client meeting. That prep session dropped from 8 hours to about 3. The quality actually improved because I spend the saved time on the parts that require human judgment — reading between the lines, talking to industry contacts, forming my own thesis about what the client actually needs. Client reports are the other big one. A typical strategy report is 15 to 25 pages. I used to spend two days writing each one. Now I spend about a day. AI handles the first draft of data-heavy sections — market overviews, competitor summaries, benchmarking tables. I write the analysis, recommendations, and anything that requires opinion or nuance. My clients have not noticed any quality change. Two of them actually commented that recent reports felt sharper.

The E-Commerce Store — Product and Marketing

My e-commerce store sells specialty desk organizers and office accessories. We carry about 340 SKUs. Every product needs a title, description, bullet points, and SEO metadata. Before AI, I had a backlog of 80 products with placeholder descriptions. I cleared that backlog in a weekend. I created a prompt template that takes the product name, material, dimensions, and key features, then generates a product description in our brand voice — practical, slightly witty, focused on the problem the product solves. Each description takes about 2 minutes to generate and edit, versus the 15 to 20 minutes I used to spend writing from scratch. Marketing emails are another area. We send a weekly newsletter to about 4,200 subscribers. I draft the email structure — what products to feature, what offer to lead with — and AI helps me write the copy. Our open rate went from 22% to 28% after I started using AI for subject line testing. I generate 10 subject line options, pick the best 2, and A/B test them. Simple but effective. Customer analysis is the hidden gem. Every month, I export our customer reviews and support tickets and ask AI to identify patterns. Last quarter it flagged that 23% of return requests mentioned the same sizing issue with one product line. We updated the product photos with a ruler for scale and returns on that line dropped by half within six weeks.

The Content Business — Writing at Scale

The content business is where AI made the biggest difference, and also where I had to be the most careful. We produce corporate training materials — facilitator guides, participant workbooks, slide decks, assessment tools. A typical training module used to take three weeks to produce. Now it takes about 10 days. Here is the workflow: I outline the learning objectives and key concepts myself. That is always human work — you cannot outsource instructional design to AI and get good results. Then I use AI to draft individual sections, generate practice scenarios, create quiz questions, and build case studies. The drafts are never publish-ready. AI-generated training content tends to be too generic and too optimistic about outcomes. Every section needs editing for specificity, accuracy, and realistic framing. My contract editor spends about the same number of hours per module as before — we just produce more modules in the same timeframe. One thing I will not use AI for in this business: writing content that pretends to be based on real experience when it is not. If a case study needs a specific industry example, I write it from my own consulting work or I hire a subject matter expert. AI can help structure and polish that content, but the core expertise has to be real.

What Didn't Work

Not everything was a win. Here are the failures I want to be honest about. AI-generated social media posts for the e-commerce store flopped. They were technically fine but had no personality. Engagement dropped 40% during the two months I tried it. I went back to writing those myself. I tried using AI to take meeting notes and generate action items from my consulting calls. The summaries missed context that mattered. A client once said something sarcastically and the AI recorded it as a genuine strategic priority. I stopped after two weeks. Automated email responses for the consulting practice were a disaster. One draft went out without my review (my fault — I set up an automation I should not have) and it recommended a timeline that was physically impossible. The client was confused. I was embarrassed. Manual review on every client communication, always.

The Real Numbers

Here is what I actually track: Hours worked per week: down from 60 to 45, with higher revenue across all three businesses. Email processing: 35 minutes versus 90 minutes per morning session. Consulting report production: 1 day versus 2 days per report. Product descriptions: 2 minutes versus 15 to 20 minutes per SKU. Training module production: 10 days versus 21 days per module. AI tool costs: about RM350 per month across Claude Pro, ChatGPT Plus, and a couple of smaller tools. The return on that RM350 is not even close to a question. The consulting practice alone saves enough hours to justify it ten times over.

What I'd Tell Another Business Owner

Start with your biggest time sink, not the most exciting use case. For me that was email processing. Not glamorous, but it freed up an hour every single day. Do not automate anything client-facing without a human review step. The risk is not worth the time saved. Track your time before and after. Gut feeling is unreliable. I thought AI saved me 30 minutes on reports. It actually saved me a full day. I also thought it saved me time on social media. It did not — it cost me engagement. And do not expect AI to replace thinking. It replaces typing, researching, drafting, and formatting. The thinking — the strategy, the judgment, the experience — that is still yours. That is the part your clients and customers actually pay for.