|
|
WEDNESDAY, OCTOBER 21, 2026
|
|
Every week we break down practical uses of AI to build smarter systems and scale faster, resulting in more profitable workflows in your business.
|
|
Sponsored By
An AI Frontier Models...an Accelerant for Your Company
|
|
Trending this Week:
OpenAI launched Dots this month, AI agents that live inside ChatGPT and the chat apps your team already uses, working in the background on your behalf. One early tester's dot noticed an unpaid invoice and got it ready to send, then waited for a yes before anything went out.
Around the same time, Meta opened its Muse agent to small businesses, wired into tools like Shopify, Stripe, Slack, and QuickBooks along with your Instagram analytics and Meta ad accounts. The pattern is hard to miss. The big platforms want AI sitting on your team, with access to your real business data, doing work while you're busy somewhere else.
That puts a new job on your plate: managing AI the way you'd manage a new hire. You decide what it can handle on its own, and you check its work before anything goes out under your name, because even the best AI writing still leaves fingerprints. MORE →
Today's Issue:
| ▶ New tools that close the loop on team and customer feedback, plus an AI marketing teammate |
| ▶ Graphite's new study: Claude Opus 5.5 sounds more human, but it still leaves AI fingerprints |
| ▶ A 15-minute scrub that takes the AI fingerprints out of your drafts |
This week is about AI joining your team, and the work of managing it once it's there.
Let's get into it. 👇
|
|
|
🚀 AI TOOLS OF THE WEEK
|
|
Tool #1
Retro
Summary
Retro is an online retrospective board from the software agency Bit Zesty. Your team adds notes in real time to celebrate wins and flag problems, then groups related notes and votes on what matters most. Notes can stay private until the facilitator reveals them, which tends to get you more honest input, and guests can join with only their name. Every insight can become an action with an owner and a due date, and facilitators can publish selected actions straight to Linear.
Why It Matters
Most retros end with a good conversation and very little follow-through. Retro ties each takeaway to a person and a deadline, so the meeting turns into work with someone's name on it, and the shared timer and parking list keep the session from eating your whole afternoon.
|
Scott's Take
"Feedback that never turns into action wastes everyone's time. I like anything that forces the question of who owns this and when it's due before people leave the room. That one habit fixes more teams than most software ever will."
|
|
|
|
Tool #2
ProductBridge
Summary
ProductBridge is an AI-native support and feedback platform for SaaS teams. It puts live chat, a help center, surveys, feedback boards, a public roadmap, and a changelog on one customer record. Its AI merges duplicate requests, sorts them into themes, and ranks them by reach and revenue impact. The support agents answer from content you've approved and hand off to a person when a question needs one, and an MCP server lets AI assistants like Claude and ChatGPT read your feedback and update the roadmap.
Why It Matters
Customer feedback usually lives in five places at once: support chats, review sites, survey replies, email, and a Slack channel somebody forgot about. When it all lands on one record, you can see which requests come from your best-paying customers, and the people who asked for a feature hear about it when it ships.
|
Scott's Take
"Your customers are already telling you what to build next. The trouble is that it's scattered everywhere. A tool that pulls it into one place and ranks it by revenue makes the roadmap conversation a lot shorter."
|
|
|
|
Tool #3
Naise AI
Summary
Naise AI pitches itself as one AI teammate that runs your social media plus your influencer and PR work, in any language. The social agent finds trending topics and turns them into on-brand posts and images, then schedules them across Instagram, TikTok, Facebook, and X. The influencer agent handles creator discovery and outreach, and the PR agent builds media lists and drafts press releases in each market's local language.
Why It Matters
Most small businesses can't pay for a full marketing team, so social gets squeezed into spare minutes and PR doesn't happen at all. Naise puts the whole lineup under one subscription, which gets especially interesting if you sell into more than one country or language.
|
Scott's Take
"I'd treat this like a junior marketer on day one. Let it find the creators and draft the posts, then put your own eyes on everything before it goes out with your brand on it. Run one campaign through it and see what holds up for your business."
|
|
|
|
Sponsored
You can't afford to be silent on ChatGPT, Gemini, or Perplexity find out why
|
|
📖 AI IN THE REAL WORLD
|
Claude Opus 5.5 Sounds More Human, but AI Writing Still Leaves Fingerprints
According to Graphite, a digital marketing agency that studies AI writing telltale signs, Anthropic's Claude Opus 5.5 writes noticeably closer to human writers than the model before it. Graphite compared ten AI models against human writers on 9,974 matched topics, and Opus 5.5's word choices came out 19% closer to the human baseline than Opus 5's. Its em-dash use fell about 99%, from 2.92 per 1,000 words to 0.015, which is now lower than the human writers in the sample.
The fingerprints are still there, though. According to the same Graphite analysis, Opus 5.5 still shows 2,548 words and phrases that turn up at least twice as often as they do in human writing, only 118 fewer than Opus 5. Some habits got stronger with the update: Opus 5.5 uses "can help you" eight times as often as Opus 5 and "is especially helpful" 12 times as often. Every model also has its own telltale signs. Graphite found Claude leans on superlatives, while OpenAI's GPT-6 Astra hedges its claims more often.
Graphite is upfront about the limits. A telltale sign on its own doesn't prove any single article was written by AI, and the human articles in the study all predate ChatGPT. Graphite's chief AI officer, Gregory Druck, pointed to one more giveaway worth knowing: AI tends to write sentences that are more uniform in length than people do.
|
Scott's Take
"Every model update moves the fingerprints around. If your team spent last year hunting down em-dashes, the newest model already dropped that habit and picked up new ones along the way. Whatever model wrote the draft, it still needs an editing pass in your own voice before your name goes on it."
|
|
|
|
Sponsored
Business credit cards built for growing companies. Apply with Chase for Business →
|
|
⚡ AI GROWTH HACK OF THE WEEK
|
The 15-Minute AI Fingerprint Scrub
The Strategy:
Your readers notice when something reads like everyone else's AI copy, even if they couldn't tell you why. Before anything goes out under your name, run it through a quick scrub aimed at the patterns this week's research points to: pet phrases and sentences that all run the same length.
1. Collect your real voice. Pull a few emails or posts you wrote yourself that sound the most like you.
2. Build your tell list. Write down the phrases you keep catching in your AI drafts. Start with the ones from the research, like, "can help you" and "is especially helpful", then add your own model's favorites as you spot them.
3. Run the scrub prompt. Paste your samples and your tell list into the prompt below along with the draft.
4. Read it out loud. Your ear catches what your eyes skip. If several sentences in a row land at the same length, break one up or combine two.
|
The AI Prompt to Steal
"Act as an editor who knows my writing voice. Here are samples of my own writing: [PASTE SAMPLES]
Here are phrases I want removed or replaced: [PASTE YOUR LIST]
Here is an AI-assisted draft: [PASTE DRAFT]
Rewrite the draft so it sounds like my samples. Replace every phrase on my list, cut any superlative that isn't backed by a specific fact, drop filler opening sentences, and vary sentence length the way I do. Keep every fact and claim the same. Then list each change you made so I can review it."
|
Why This Works:
Graphite's research shows each model has its own habits, and those habits shift with every update. A telltale sign list you build from your own drafts tracks the model you actually use, so it stays current when the model changes. Anchoring the rewrite to your real samples gives the AI a concrete voice to aim for, and the change list keeps you making the final call.
|
|
Wrapping Up
Dots and Muse grabbed the headlines this week, and they show where the big platforms are headed. AI is moving into your business as a teammate with access to your real data, and that makes your job look more like managing than doing.
The smaller tools fit the same picture. Retro turns team feedback into actions with owners attached. ProductBridge pulls customer requests into one place and ranks them, and Naise AI picks up the marketing work most small teams never get to.
Graphite's research is the reminder that comes with all of it. AI keeps getting better at sounding like us, and it still leaves fingerprints a careful reader can pick up. A new teammate still needs someone checking the work, and that someone is you.
Run this week's scrub on the next draft you're about to send. Fifteen minutes is a small price for something that sounds like your business actually wrote it.
Keep profiting,
|
|
Keep profiting,
Scott Hall
AI Profit Letter
|
|
📕 Operational AI Profit Wave
|
.jpg)
|
|
|
|
|
You're receiving this because you subscribed to the AI Profit Letter.
Better Online Info LLC | AI Profit Letter | 2150 NJ-35, Suite 250, Sea Girt, NJ 08750
|
|
