Last Wednesday night, I nearly missed a collaboration
I was sitting in my Hangzhou studio with client messages across WeChat, email, and voice notes. A potential client surnamed Chen had asked for a quote three days earlier, but all I remembered was “let’s talk later.” I spent two hours digging through the records that night, because I had to get this moving rather than wait for a system to rescue me.
My approach: give AI the organizer job
I copied client conversations into AI and asked it to extract names, needs, agreements, and next steps, then output a follow-up table. It wasn’t meant to chat with clients for me; it was there to stop important details from scattering. My friend Sister Lin, a brand consultant in Chengdu, feeds the same set of prompts 5–8 client messages each week, generates a to-do list, and uses calendar reminders for herself. She also got stuck: the first time she asked AI to “summarize this,” she got vague sentences. It became useful only after she added, “Keep times, prices, and commitments; mark anything uncertain.”
Today’s replica: ¥0, about 30 minutes, almost no technical barrier
Money: The free allowance on a common conversational AI is enough, so the budget is ¥0; if you already pay for a plan, use that. Time: About 30 minutes for the first setup. Technical barrier: Copying and pasting text is all it takes—no coding. First step: Open the AI chat tool you already use, start a new conversation, and paste this: “From the client communication below, extract the name, need, date, price, commitment, and next step; mark missing information as ‘to be confirmed’ and do not fill in the blanks yourself.” Then paste one real chat with identifying details removed and test it.
If I’m busy right now, I start with one small step
If I’m just getting started, I’d spend five minutes a day copying the day’s most important client commitment into a list. There’s no need to build a full workflow yet, and it’s fine not to try today—just preserve the original conversations. With 1–2 steady clients, I’d fix three columns: client, need, and next action, then have AI flag anyone I missed following up with each week. As I scale, I’d add source, amount, owner, and deadline, plus spot checks. The more people involved, the more manual review I need; AI isn’t the person who closes deals automatically. Not everyone needs this setup. If my communication is already organized, a simple spreadsheet may take less effort. Which client message would I test first?