One project per sale, one step per chat

I run each sale in its own Claude project, with instructions so every chat inside it knows the sale. Then I work through my Flash Sale AI Command Centre (an AI tool I created for myself and my Aligned AI Lab members), one step per chat, and each step saves a file back into the project.

1. Claude interviews me and shortlists what could go on sale. It already knows my offers and prices, so it fills in the blanks.
2. It asks me the decisions one at a time (how many days, which deal goes when) then locks in a day-by-day plan.
3. We agree one repeating email layout, then it drafts every email to that shape. I edit them and send them to Kit as drafts.
4. It builds the sale page from the plan. The page is programmed to switch to the next deal by itself each day.
5. It turns everything into an action plan with a calendar, so I'm not doing it all the night before.

I still choose the offers, edit every email, set the deadlines on my checkouts, and press publish.

It starts from what I actually sell

I trained Claude on my business first, creating what I call my "Business Brain": my voice, my offers and prices, my brand visuals. So it starts from what I actually sell and how I actually talk.

Then I edit every email by talking my changes out loud, and run my sales copy through my Sales Copy Sharpener skill (available inside the Aligned AI Lab), which checks whether it would actually make someone buy, not just whether it sounds "nice".

“The final word on the offers and the emails. Claude can shortlist what goes on sale and draft every email, but I choose what my people are offered and I edit every email before it goes. I always say Claude is my helper, but I'm the visionary behind the business. I know my offers and clients best.”

What Ruth would never hand over

“I don't let AI decide what I charge for my offers. I might use Claude for a little market research or to compare and contrast what similar offers cost, but I'm always the one making the pricing decision for every offer in both of my businesses.”

What Destini would never hand over

Notes in, a designed draft waiting in Kit

My setup is a Claude skill plus the Kit and Airtable connectors. The skill holds my full issue structure, my design system as Kit-safe HTML, and my content rules, like no discount codes in the newsletter and no em dashes ever.

Each week I give Claude the real moment from my business I want to open with, and it pulls the rest from an Airtable table where I drop things worth sharing all week.

Claude drafts all four sections, writes the "everything mentioned today" recap, builds the HTML, and saves it as a draft broadcast straight into Kit. I still pick the story, rewrite anything that doesn't feel right, check every link, and hit schedule myself.

Every correction becomes a permanent rule

The biggest difference was writing my rules into the skill every time I corrected something, instead of fixing the same mistake in chat each week. When Claude named a coupon code I don't give away freely, or described my membership as including everything I've ever made, that became a permanent rule.

The skill is now a record of my judgment, so each issue starts where the last correction left off.

“The story. Claude can shape it, but the moment itself has to come from my real week, like a launch that flopped or something my daughters said at dinner. That's the part readers reply to, and replies are what keep my emails landing in the inbox. A made-up story would break that trust instantly.”

What Monica would never hand over

“I'd never hand over the sending of the outreach messages (whether that's by email or DMs). Claude can find the person, gather the history, and draft the first note. The listening, the judgment about whether someone is ready for a next step, and the relationship itself stay with me, because that's where trust gets built.”

What Lisa would never hand over

Two skills, one scheduled task, and she still hits send

1. I pick the people. That part is mine. The list lives in an Airtable table, one row per person.
2. Claude finds the right email address for each one, starting with my own inbox, and notes where it found it.
3. My pitch writer skill, Scout, drafts the pitch. It learned from pitches I've sent that got a yes. Everyone gets the same core email plus two lines written only for them: an opener and one personal line.
4. I read every one and fix what's off. Then I tick a box and it goes out from my own Gmail. Claude never sends on its own.
5. A scheduled task reads my inbox every morning, sorts the replies (yes, no, maybe, question), updates the tracker and drafts my answer inside the same thread.
6. Anyone who hasn't replied after about five days gets one nudge. After that I let it go and ask again next round. (I'd rather have a fast no than a slow maybe.)

Every number has to come from her own records

It never starts from a blank page. My pitch writer learned from real pitches I wrote myself, the ones that got a yes, so it borrows my moves: open on something true about them, put my proof in the first two sentences, offer two or three ways to say yes, and end on a question.

It also has two rules it can't break. Every number has to come from my own records (it can't make one up), the main point of the pitch is always "what's in it for them" and every pitch waits as a draft until I've read it.

“Choosing who to pitch, and hitting send. Claude drafts every pitch and every reply, and then they wait for me. A pitch is me asking a real person to lend me their audience, so a real person reads it first. What comes after the yes (showing up, promoting, saying thank you) stays mine too.”

What Kate would never hand over

“Replies and conversations. When someone comments on a post or sends me a message, that's a real person who took the time to reach out, and they deserve to hear from me. Claude can help me show up with ideas, and the relationships that grow from those posts are mine to tend.”

What Tara would never hand over

One question at a time, answered by voice note

My setup is a custom skill. It holds my fixed newsletter template (3 Things On My Mind, 2 Business Updates, 1 Google Tip), my voice rules, the phrases I never want to see, and examples of my real subject lines, all built from 24 newsletters I had already sent. Here's how a Thursday goes:

1. I open Claude after work and start the skill.
2. Claude asks me one question at a time, starting with 'What's one thing that happened this week that you keep coming back to?' I answer by voice note. It never hands me a list of prompts, which matters, because a list is exactly where I used to stall.
3. It collects my three reflections first, then asks what's happening in the business, then asks whether I have a Google tip in mind. If I'm stuck on the tip, it offers two options based on what I already said, and no more than two.
4. Claude writes the full email into my template and gives me five subject lines with matching preview text. I pick one by number. It asks what feels too long or too polished, and fixes only what I flag.
5. What I still do myself: the stories and opinions are all mine, I choose the subject line, I do the final read, and I paste it into Kit and schedule it for Friday morning.

Every story comes out of her own mouth

The biggest difference was setting it up so Claude never comes up with the content. It interviews me. Every story, opinion, and Google tip in my newsletter comes out of my own mouth in a voice note, so the email is made of my words before Claude writes a single sentence.

Its job is to shape what I said, and I never ask it to invent something for me to say.

That's also why it still sounds like me. I talk more naturally than I write, so a voice note already has my phrasing, my humor, and my tired-on-a-Thursday honesty in it. And at the end it asks me what feels 'too polished,' which is my cue to catch anything that sounds like AI wrote it.

“The talking. I'd never let Claude skip the interview and write from a topic alone. Voice noting my answers is how it gets the full context: what happened, why it mattered to me, and what I want my readers to take from it. Without that, it's guessing at my intention, and my readers would feel the difference.”

What Adrienne would never hand over

Six steps from voice note to scheduled in Kit

My setup is a Claude skill I built from my own sales email lessons, my subject line method, and real emails I've sent.

1. I record a voice note or type a messy brain dump about what I'm launching, and add the offer name, price, link, and deadline.
2. Claude tells me the main problem it heard and the best direction for the email.
3. It gives me subject lines grouped by angle, with one recommended pick, and I choose one.
4. It writes the email from my own words, with preview text options, the link near the top, and a PS and PPS, and checks that prices, dates, and links match.
5. I go through it once and dictate the email again the way I would want it, if I'm not happy with the language or any parts of it.
6. We have a final draft and a final edit, then I add any live numbers and schedule it in Kit.

The skill trims what she said and keeps her tone

The biggest difference was telling Claude to keep my own words from the brain dump instead of rewriting them. My best copy has always come from how I speak, so the skill shapes and trims what I said rather than inventing new lines, and it learns my tone and emoji style from my real emails.

And because I dictate my final changes the same way, the email ends up sounding like me talking to one person. It also never makes up numbers, results, or testimonials, which matters to me because honest urgency is how I sell.

“Starting from scratch. Even though Claude can give me ideas, I really like the first draft to be mine, even if it's just a messy voice note. That first draft holds my words, my flow, and the stories I want to tell, and that's what makes the final email sound like me.”

What Asmita would never hand over

“In general, I write all my emails and the bulk of my copy. Although Claude can help brainstorm angles, copy and more, I prefer to maintain connection with my audience by writing the actual contents myself. It's what keeps my readers loyal, open and click rates high, and continues to build trust.”

What Naima would never hand over

“Asking a client for a testimonial, and deciding which client story to tell publicly. Evie can find the proof, but the ask is a relationship moment, and so is choosing whose story goes on my sales page. Those stay with me, because trust is the whole point of the proof.”

What Jodi would never hand over

A full share kit in about 10 minutes

It's a skill in Claude, not a one-off chat. I give it the offer name, the link, the price and who it's for. Claude reads my sales page and my brand details, then writes share emails with subject line options, a few social posts, a short newsletter blurb and image ideas.

It saves everything so it shows up on my affiliate page, ready for anyone to grab.

I still read every piece and tweak anything that doesn't sound right. It works the same for an offer I've had for years or something I'm launching next week.

Written so whoever sends it sounds like themselves

I set up a writing style skill once, with how I actually talk, words I use, words I never use, and real emails I've sent. Every kit is built on that. And the share copy is written so whoever sends it sounds like themselves recommending something they use, not like they're reading my ad.

“The relationships. Claude can write the share copy, but the personal thank-you when someone recommends me, and noticing who's already talking about my work, stays with me. People share things because they trust you, and that's not something I want sent on my behalf.”

What Steph would never hand over

One line to Claude, three to five Threads posts back

My setup is a skill I built called Sent to Threaded, plus my brand voice guide saved as its own skill, with my email platform connected to Claude. I type something like "pull my last sales email and turn it into Threads." Claude finds the email in my email platform, checks that the offer, link, and any deadline are still live, then gives me 3-5 posts.

Each one is labeled with its angle (I have a framework I follow that's built into the skill), whether it has a direct or soft CTA, a character count, and a suggested topic tag. I pick my favorites, make my tweaks, and do the posting myself. I've found using schedulers just... complicates things. I can copy + paste + schedule in 15 seconds flat, so sometimes the simple way is the way.

Nothing goes in a post that wasn't in the email

The biggest key was insisting that Claude work from what's actually in the email that I carefully wrote and isn't allowed to invent numbers, results, or urgency, so every post stays true to what I already said. Since my brand voice skill is connected in Claude, it'll add little tweaks and texture, but the output still sounds like me, written + formatted the way I would.

“The finished, polished version of the email I send my list and any personal stories I include in it. Claude can find the angle in an email I already wrote, but the confessions and "here's what I got wrong" moments that my audience loves to read have to be lived by me first. If I asked AI to invent a story, my audience would feel it, and the whole thing goes down the drain fast.”

What Dama would never hand over

“The strategy calls and the final edits. Claude holds the strategy with me, but deciding what we're really offering and why it matters right now is ours. And the last read-through is always me, because I'm the one who can tell when something sounds like us and when it just sounds polished.”

What Kinsey & Joyce would never hand over

Coaching calls in, a week of reviewed drafts out

My setup is a set of Claude skills plus a private dashboard I call my Control Room. Here’s the flow:

1. I coach as usual. Transcripts from my client calls, podcast and videos come in automatically. I don’t have to remember to save anything.
2. Claude finds the content in what I said. My skills look for three kinds of moments: a question I answered clearly, a story I told, and a strong opinion. Each one becomes a social post, carousel, email or video script, using my actual phrases wherever possible.
3. Everything lands in one review queue in my dashboard. Each card shows the finished draft, the exact quote and timestamp it came from, and the one thing I need to check, like a claim, a link or a name. Only finished drafts get in. Ideas stay out, so the queue never turns into another to-do list.
4. I review on the card. I edit the draft right there, then choose Approve & schedule, Redo or Later. Redo requires a note (too generic, wrong hook, missing the story), so the rewrite comes back better.
5. Approved pieces move to the schedule queue, and they’re only marked published once there’s proof they went live. The dashboard’s home page shows me what needs my review, what’s scheduled and what’s already handled.
6. What stays with me: the coaching, the final yes, and keeping clients’ private details out. Only my words become content.

You don’t need my setup to start. Paste one call transcript into Claude, ask for review cards in that format, and keep the approved ones in a single doc. That doc is your first dashboard.

Claude shapes what she already said to a real client

It starts from my actual words. Claude isn’t inventing content and trying to sound like me; it’s shaping things I already said to a real person with a real problem, and it keeps my exact lines wherever it can. That’s why it sounds like me.

On top of that, whenever I edit or Redo a piece, I tag what was off (voice, hook, proof or structure), and that correction becomes a permanent rule for the skill that wrote it. I stopped fixing the same line every week.

“The original ideas. I create the concepts, frameworks and opinions, mostly while coaching, when a client’s question pulls the real answer out of me. Claude repurposes that core content beautifully, but when I’ve asked it to come up with ideas for me, they sound like everyone else’s. My thinking is the product. Claude just helps it travel further.”

What Faith would never hand over

From a booked call to a one-page game plan

My setup is a custom Skill built by me for sales call prep, and run by my {AI} Sales Director employee. Here’s the flow:

1. When I see an Intro Call booked, I ask my Sales Director to grab the lead’s application information from my calendar software.
2. My Sales Director runs the Sales Call Prep Skill, which researches the lead and evaluates how well they fit my ideal client profile and my offers.
3. From there, it creates a one-sheet for me: who they are, a score for how likely they are to convert, their likely objections and needs, and a script outline to follow.
4. I read it, fix anything that’s off, and add my own gut read.

Every AI employee reads her Brand DNA Skill first

Whenever my AI employees in Claude do work for me, they start by reading my Brand DNA Skill. It’s an in-depth reference Skill that holds my offer details, my ideal client information, and my brand and style guidelines. That’s why everything they hand back sounds like me, fits my business, and I don’t have to fix generic outputs after the fact.

The Sales Call Prep Skill also includes a 5-part call outline that I follow for my sales conversations. So the prep doesn’t just tell me about the lead. It gives me a plan for how I like to run the call.

“I still do the call, the relationship, and the final decision on which offer fits. My Sales Director does the homework. I do the human part.”

What Wendy would never hand over

“The yes. Deciding who I partner with, and the relationship after the first email, stays with me. Every collaboration puts my name next to someone else's in front of my list, and that's my reputation. Claude can find the room and draft the knock on the door, but I choose which doors.”

What Alisa would never hand over

An editorial team that never forgets a step

My blog runs on a system of AI routines that handle the research, drafting, checking and measuring, while I stay in charge of what goes live. Each week it finds what my audience is searching for, drafts a post in my voice from a brief, and puts the draft through a series of editorial checks before it reaches me. I edit it, approve it, and publish it, and the system never publishes anything without my say-so.

Several layers, not one setting

It sounds like me because several layers work together. The skill interviews me for my real stories and numbers first, then writes from a voice system I've built for my business, backed by examples of what sounds like me and what sounds like flat AI.

A no-slop pass and a cold-read check catch anything that reads like AI. I fact-check and edit before approving, and my edits feed back into the rules so each draft needs fewer fixes than the last. Nothing goes live without my say-so.

“Publishing and fact-checking. Claude writes the draft, but I read every line, check every claim, and approve it myself. A post speaks for my business in public, and the people reading it trust me to have stood behind what it says, so that part stays with me.”

What Sherise would never hand over

Upload, say "edit", done

The first thing I did was set up a brand voice and audience skill, so Claude knew who my audience was and how I normally talk. That informs it when it's editing. Then I created a brand board in Claude Design. I am not good at design at all, so using Claude Design to work out what I was really looking for helped a ton.

After that I record a video raw on my phone, just going through the things I want to say. Then I upload it to Claude and have it use the video editing skill I created to edit it how I want. I don't have to ask it to cut things or tell it what graphics or branding to use, because it already knows all of that from what I've given it before. I can just upload, say "edit," and it takes it from there.

A rubric it checks before I see anything

Having the skills in place is what makes it still sound, feel and behave like me. Because of the skills I don't have to re-explain every time. I don't have to say, "Okay, you need to do this or that." It just reviews that automatically.

We also have a rubric in place so that, before it turns anything in to me, it looks through the rubric to make sure it's behaving like it's supposed to and not missing anything. It does the checking before it gets to me, so when I see it, I'm not seeing the first draft. I'm seeing the finished product.

“The human connection part: understanding my audience, talking with them, hearing their concerns, fears, excitement, wins. I WANT that connection and I want to give it. I think it's a beautiful thing to connect with incredible people and I don't want to miss out on that even if it means being a little less efficient sometimes.”

What Brittany would never hand over