Sales Director · AI Implementation

Braydon Hart

I build AI systems that do real work, from a sales assistant my team uses daily to an AI crew that runs an Amazon store, and I train people until the tools actually get used.

See my projectsWhat I'm building now
Metro Detroit · Open to remote LinkedIn braydonhartgtb@gmail.com
$30.6Mlifetime loan volume
434loans closed by the team I lead since 2024
#1closer at my company in 2023
2reps I trained who were promoted to leaders
Top 1%UWM award, 2024 and 2025
30–60minutes saved per rep, per day, with my AI sales assistant (estimated)

About me

From top closer to AI builder

I started as a loan officer in 2022 and became the top closer at my company in 2023. That led to a promotion to Director, where I've led a team of 9 loan officers since 2024.

Leading the team taught me that my real strength is teaching. I learned to coach people to own their work instead of doing it for them, and two reps I trained have become leaders themselves.

Along the way I started using AI to remove the busywork slowing my team down. I found that building the tool is only half the job. The other half is getting people to actually use it. That's the work I want to do full time.

Looking for: AI implementation, AI enablement, or sales enablement roles. Remote or Metro Detroit.

How I work

Four steps, every project

  1. Find the friction

    Watch where reps guess, repeat themselves, or lose deals between steps.

  2. Build the tool

    Use AI to build something small that removes that step.

  3. Train the team

    Teach people to own the tool, with clear expectations, so it doesn't depend on me.

  4. Simplify after real use

    Cut whatever people skip. A simple tool that gets used beats a detailed one that doesn't.

Selected work

Projects

Built

  • Automated Sales System (n8n)Running
  • Stitch BenchLive AI
  • Azoki MarketLive on Amazon
  • LO Command StationPrototype
  • Sales Cheat CodeTeam trained
  • 4 team tools (dashboards, proposals, tracker, templates)In use

In progress

  • Sales Cheat Code LivePilot proposal
  • Agent StationPrototype built
  • GradeWorthItBuilt, launching

Automated Sales System

Built in n8n · running
Claude+OpenAI+Google Sheets+Gmail→n8n workflows
1. Chat coachPaste a client text or call notes, get a ready to send reply, a backup, the next step, and a follow up plan.
2. Morning follow upsEvery weekday at 8am it finds who's due, drafts a text for each lead, and emails the list hottest first.
3. Call listeningUpload a call recording; it transcribes the call, writes a recap and follow up text, and logs it to the sheet.
4. ScoreboardA live, gamified page with XP, levels, streaks, a daily quest, and one click follow up logging.

Chat coach

Client texted: rates are too high right now, I'll wait.
Read"I'll wait" is usually a soft brush off. They haven't said no to their goal. Warm lead.
Send thisTotally get it, rates are on a lot of minds. The bigger question is what you want your home to do for you, like a lower payment or cash out. Want a quick no obligation 10 min review? Does today at 4 or tomorrow at 10 work better?

8am follow up email

MorganWarm · attempt 3
Since your spouse needs to be on the call, let's find a time you're both free so nothing gets missed. Does tonight at 7 or tomorrow at 12 work better for both of you?
CaseyCold · attempt 5
I've reached out a few times and haven't heard back, so I don't want to keep bugging you. Should I close out your file for now, or do you want to take a quick look at your home options?

Real output from test runs with sample leads. Notice the angle changes as attempts go up, and texts never quote rates.

Sales Cheat Code scoreboard showing level, daily quest, streak, follow ups due with Log It buttons, pipeline, recent calls and badges
The live scoreboard, shown with sample data.

Problem

My first assistant still needed a rep to stop and ask it something. Follow ups, call notes, and tracking still depended on memory and manual updates.

What I built

Four connected workflows in n8n that run on their own: scheduled follow up emails, automatic call analysis, a chat coach, and a scoreboard that turns follow ups into a game.

Result

Tested end to end with sample data. Compliance rules are built in: texts avoid restricted terms, and sensitive numbers never reach the recap or the sheet.

Stitch Bench

Live AI · working
Photo or idea→AI designer drafts→Stitch inspector + art director review→Designer revises→Machine checks→Machine files
Stitch Bench workbench showing machine and garment settings, the photo upload panel, and an embroidered sea serpent design in a hoop
Stitch Bench with its built in example design.

Problem

Turning a photo into a design an embroidery machine can actually sew takes design skill plus technical stitch knowledge most people don't have.

What I built

A multi agent AI design team. A designer drafts original art, two AI reviewers critique it, the designer revises, then the tool converts it to stitches and runs automatic checks on the final machine files.

Result

Produces validated VP3 and DST machine files, a thread list, a sew out guide, and cost and price per piece. The review loop catches problems before anything gets stitched.

Azoki Market

Live on Amazon
Public domain art→AI crew builds listings→Printify→Amazon→Crew checks listings + reports sales

Problem

I ran a profitable blanket shop on Etsy from 2022 to 2024 and know firsthand how much repetitive work a store takes: designing products, writing listings, publishing, and checking what sells.

What I built

An AI crew that runs a woven blanket store using classic public domain art like Hokusai's Great Wave and 1938 national park posters. It builds listings in Printify, publishes them to Amazon, confirms they went live, and reports sales, traffic, and profit to a live dashboard.

Result

Launched in October 2026 with 12 designs live on Amazon. Anything that spends money or needs a decision goes to me for approval.

LO Command Station

Live AI · prototype

Problem

Loan officers need daily coaching, ready to use scripts, and a plan for the day, but borrower data can't go into outside AI tools.

What I built

A gamified command center where an AI lead (ATLAS) reads the loan officer's weekly goals and funnel, picks a focus, and sends missions to an AI crew that drafts templates and coaching. The loan officer approves everything that goes out.

Result

Privacy by design: no borrower names or loan details are ever entered. Templates use placeholders, and the funnel is tracked as counts only.

Sales Cheat Code

Team trained · rebuilt in n8n
Loan officer call→AI listens→Suggests next step→Follow up reminder + reply ideas

Problem

Reps were guessing when to follow up and what to say back to client texts, so leads went cold and every rep handled the same situation differently.

What I built

An AI assistant that listens to calls and tells the loan officer what to do next, reminds them when to follow up, and recommends how to answer a given text.

Result

Trained my team of 9 on it, and all 9 use it daily. It made follow ups an estimated 20% faster and saves 30 to 60 minutes per rep per day.

More tools I've built for my team

Team Funnel DashboardsTrack all six stages from dials to underwriting per rep, so weekly coaching targets each rep's exact drop off point.
Automated Client ProposalsTurn a loan scenario into a branded HTML and PDF proposal with payment comparisons and disclosures, with a preview step before anything goes out.
Pipeline TrackerAn Excel tracker I cut down from too many columns to name, phone, stage, and notes so the team would actually use it.
Compliant Outreach TemplatesSMS templates rewritten with AI around texting platform rules, so follow ups get delivered instead of flagged.

Currently building

In progress

Sales Cheat Code Live

Pilot proposal

What it is

Connecting the automated system to the tools my team already uses, so Dialpad's call transcripts and Bonzo texts turn into recaps and reply drafts automatically.

Where it stands

Designed with compliance first: sensitive numbers filtered out before the AI sees anything, a licensed loan officer approves every message, and company owned accounts only.

Next step

Pitch a 30 day pilot with my team to leadership.

Agent Station

Prototype built

What it is

A multi agent AI system: a "governor" model makes the big decisions and directs worker bots that run the tasks, all shown as a gamified control room.

Where it stands

Two working prototypes: Card Shop Station, where a live AI crew builds and markets digital products for collectors with an approval queue, and Pipeline Station, a simulation for testing the game loop before real AI is connected.

Next step

Connect Card Shop Station to a live store, the same way the Azoki Market crew already runs.

GradeWorthIt

Built, launching

What it is

A free web calculator that shows collectors what grading a card actually pays after every fee, based on sold prices, their odds of each grade, and grading and selling costs.

Where it stands

The calculator is built and working. The domain gradeworthit.com is registered and set up on Cloudflare.

Next step

Publish it on gradeworthit.com to bring collectors in without paid ads.

Before AI

I've always built and sold things

Long before AI, I was teaching myself whatever a project needed: product design, packaging, factories, online stores and ads.

Vinyl collectible figure

Designed, made and sold · 2022
My sketch→Box design→Overseas factory→Sample + changes→Final run→My website + Instagram ads→Collectors' shelves
Hand drawn character sketch of the vinyl figure in a cream tee and light blue jeans holding an orange
My first design for the figure.
The finished vinyl figure inside its window box with Azoki Market branding
The finished figure in the box I designed.
A collector's photo of the boxed figure on their shelf
Collectors posted it on Instagram after it arrived.
Flat die line layout of the window box packaging with Azoki Market branding and an illustration of the character
The packaging I designed, laid out flat for the factory's printer.

The goal

Turn a character I drew into a real collectible, with no background in toys, manufacturing or packaging, and no AI tools to lean on in 2022.

What I did

Designed the figure and its window box, found an overseas factory, reviewed a physical sample, made changes, and approved the final production run.

Result

Sold it on my own website under my Azoki Market brand, used Instagram ads to get the word out, and collectors shared it on their shelves.

"The hardest part was figuring out how to make something from scratch, and the problem solving it took to learn every new skill along the way."

Custom blanket and tapestry shop · 2022 to 2024Built and ran a profitable Etsy shop selling my own custom blanket and tapestry designs. It's the business my AI crew now runs on Amazon as Azoki Market.

Tools

What I build with

ClaudeChatGPTOpenAI APIn8nGoogle SheetsGmailMulti agent systemsPrintifyAmazon SellerPrompt designHTML to PDF automationExcelSalesforceBonzoDialpad

Certifications

AI Fluency: Framework & Foundations · AnthropicClaude 101 · AnthropicLicensed Mortgage Loan Originator · NMLS #1925352

Let's talk

I'm looking for an AI implementation or enablement role where I can help a team adopt AI and see real results. The fastest way to reach me is LinkedIn or email.

LinkedIn braydonhartgtb@gmail.com