A Live, Hands-On Course
December 9 & 10, 2026
9:30–11:30 a.m. & 1–3 p.m. ET each day
Every time you open an AI chat, it starts from zero. You explain who you are, what you're working on and how you write, again.
Leave with a chief of staff that knows you and your work.
At the end of Day 2 you have an AI assistant, set up in Claude, that already knows your work, plus a second AI that checks its work.
Your chief of staff is the name we give that assistant. Your control room is the setup where the assistant and its reviewer live.
Two days, live online, with an instructor. Built for beginners. You build it on your own laptop, for your own work and life.
Reserve Your Seat — $495On Day 1 you build the assistant. It reads your rules, your people, and what matters right now, and it carries them from one session to the next. On Day 2 you build the second AI that checks its work.
It already knows what's on your calendar, what you promised people, and how you like things written. You start where you left off. If you take on a small project for someone else, it knows that project too.
A separate reviewer checks the draft against your source documents before it reaches you. A second set of eyes on the work.
You staff it for what you need, whether it's a single digital chief of staff or an entire workforce.
Built for beginners. If you can describe what you want in plain English, you can build it. Each topic is taught, shown live, then built by you, so you leave with something that works on your own laptop.
The technical name is “admin rights.” Most personal laptops have them.
You sign up with Anthropic, not with us. Claude Code, the tool we build with, is included in the paid plans.
Use sample material, not anyone else's private information.
On Day 1 you see where your information goes and set the privacy choices on your own account. You decide what goes in.
AI can get things wrong. On Day 2 you build a reviewer, a routine you run: a second AI reads the first one's draft against your own documents and flags what doesn't match. You make the final call.
On Day 1 you set the rules for what it can do on its own and what it has to ask you first. On Day 2 you set up a backup of what you've built so you can restore it.
You work in a terminal window, and you type plain-English instructions to Claude. An optional tech-support session before class gets your laptop ready, and optional lab time comes before each session.
These run in the instructor's own control room every week. Each is made the way you'll make yours, from skills and agents. A skill is a saved set of steps your assistant follows. An agent is an assistant that does a job on its own.
A daily briefing waiting when you sit down — your calendar, what matters, what to watch. You build this one on Day 2.
Before a meeting that matters, it pulls what you already know about the person and what's at stake, and hands back what to land, what to avoid, and your edge.
At the end of a working session, it scans back for anything worth keeping — a decision, a new contact, a commitment — and asks before it saves. It does the remembering for you.
A standing panel of advisors with different lenses — the builder, the investor, the performance scientist, the meaning-maker — on call anytime a decision is big enough to want a second opinion.
Show it what's on your screen — a stuck printer dialog, a confusing form, a setting you can't find — and it sees what you see.
Today's session arrives with what you did last time and your next target. Text back what you actually did, in shorthand, and it's logged.
Wednesday, December 9 & Thursday, December 10, 2026. Four two-hour live virtual sessions, 9:30–11:30 a.m. and 1:00–3:00 p.m. ET each day, eight hours of instruction in all. Each module is taught, demonstrated live, then built by you.
By the end, you have a chief of staff that knows you, a team around it that checks its work, and the templates to keep building. Add the next skill or agent the same way and your setup grows with your work, from a single assistant to an agentic workforce.
Extra hands-on lab time on both days, right before each main session: 9:00–9:30 a.m. and 12:30–1:00 p.m. ET. Completely optional and on top of the eight core hours — drop in for more build time, or skip them entirely.
You've seen this work with people: one person writes it, another reads it before it goes out, and you make the final call. Day 2 builds that around your chief of staff.
A reviewer is a second set of eyes on the work before you rely on it. Sign-off stays with you.
One proof the method works: the skills and agents you'll learn, applied to a store. Yours can be smaller and closer to home.
Before any real data was involved, the team built a working inventory dashboard on made-up orders and products. That practice set stayed on as a test: after any change, the same numbers have to come out.
Once the store's Shopify exports arrived, the dashboard was rebuilt to read them, and a check compares its totals to the source files. When the tool once recommended reordering gift cards, the fix went into all four places that recommendation is calculated.
A Shopify configuration skill looks at the store the way a visitor sees it, pulls Shopify's current help pages for every setting it touches, and writes the steps with a link at each one. For any change that would make a store visible or buyable, it locks purchasing first and checks as a logged-out visitor before calling the change done.
I recently completed the Heflin Group's "Building Your AI Digital Workforce" course. It was outstanding! Whitney's knowledge and speaking style made it very easy to follow. We reviewed some foundational material then dove into the mechanics of how to build a digital workforce. We covered the things you need to get started -- from initial set up, to how to build agents, to skills and backup strategies and everything in between. There were even some templates and practical exercises to make sure we were learning and able to do it on our own. If you want guidance on setting up your own workforce, this is the place to get it. Thanks Whitney!
Founder and Fractional CFO, Heflin Group LLC · Navy veteran
Whitney spent 20+ years in risk management, audit, and finance inside one of the country's largest health insurers, and she co-owned and ran a yoga studio. As a CPA, she helps business owners and leaders solve business problems, and her work crosses finance, operations and technology.
Today she runs Heflin Group LLC and directs a digital workforce, a team of AI agents she built that does the work. She built it while training under one of the world's leading AI authorities, and an app she built was featured to the whole cohort in the program's final session. This course is her handing you the shortcut.
Left brain, right brain, all of it: she'll talk risk models and yoga flows in the same breath, she knits, and she lives in Louisville with three English Cream Golden Retrievers who remain thoroughly unimpressed by artificial intelligence.
Cooper is Whitney's chief of staff. Cooper works with Booker, Knox and nine directors. A director can have sidekicks: smaller agents that each do one narrow job for that director. Together they are the agentic workforce Whitney built and runs. In the course you build a small version of this: one chief of staff and one reviewer.
This team is the product of months of work. When an agent gave a made-up answer, Whitney found the cause and added a rule or a check. The team also checks itself on a schedule, and looks at how others run agents before building anything new.
Select any name to see its purpose, the model it runs on and what it is accountable for.
As of October 2026. Models change. Opus, Sonnet and Haiku are families of AI models from Anthropic.
These run on a schedule. The agents draft, route and check, and Whitney makes the final call.
Every night
Runs as Booker. Keeps the workforce's shared memory in order.
Every night
Notes which agents have been active.
Every morning
Whitney gets a briefing on the day ahead, a status note for each active project and a back-office list.
Every morning
Confirms the morning briefing went out.
Every day
Sorts open loops into what Whitney owes and what she is waiting on.
Regularly
The workforce's files are backed up regularly.
Sunday evenings
Runs as Cooper. Tidies the week's lists.
Every day
Runs the annual-filing-reminder. It speaks up only when a filing is due.
It fits you if you work on a computer and you're tired of explaining yourself to an AI from scratch every time. The course is built for beginners: if you can describe what you want in plain English, you can build it. You choose a real task from your own work or life and build around it.
You don't need to. This course is built on and for your own personal laptop and your own work, and there is plenty to build there: a daily briefing waiting when you sit down, preparation before a conversation that matters, and a record of what is worth keeping from each working session. Anything that belongs to your employer, such as their files, their clients' information or their laptop, falls under your employer's rules.
No. It's for anyone who works on a computer and wants an assistant that already knows their work and their week. You bring a real task of your own and build around it.
No. You don't need any AI experience, and the course starts with the basics. You build in Claude, so you set up a Claude account first, and the setup checklist walks you through it.
You choose the task, so you build it around your own work. The method is the same for any task: name a recurring job, tell the assistant what it can see and what the rules are, and test it.
No. You don't write code in this course. You work in a terminal window, and you type plain-English instructions to Claude. If you get stuck on setup, an optional tech-support session before class is open to you, and optional lab time before each session gives you more hands-on time.
Day 1 starts from a blank page: you write the one file your assistant reads at the start of every session. Then you pick one thing you do every week, and on Day 2 you turn it into your first skill, a reusable set of steps. You leave with templates to keep building and a simple plan for what to build next. An optional tech-support session before class gets your laptop ready.
AI can get things wrong. On Day 2 you build a separate reviewer: a second AI that reads the first one's draft against your own documents and flags what doesn't match. It's a routine you run, and sign-off stays with you.
What you type, and what you ask it to read, is sent to Anthropic to produce the answer. You control the setting on your own account that decides whether Anthropic may use your chats and coding sessions to improve future models, and you choose what you give it. On Day 1 Whitney walks you through where your information goes, live, and you set your privacy choices. Then you decide what goes in.
The details are on Anthropic's own pages: Claude Code data usage and model training and your data.
It can edit files and run commands on your computer, and you decide how much it does without asking. On Day 1 you set those rules live with the instructor: what it can do on its own and what it has to ask you first. On Day 2 you set up a backup of what you've built so you can restore it.
Anthropic's page: Claude Code security.
Use practice or sample material in the course, not anyone else's private information. Beyond the course, you decide what you give it, and anything that belongs to someone else, such as an employer or a client, follows their rules. Heflin Group can't promise how any tool handles your information; Anthropic's pages describe what it does.
An AI assistant, set up in Claude, that already knows your work, plus a second AI that checks its work. You also leave with templates to keep building, a way to back up what you built, and a simple plan for what to build next.
Wednesday and Thursday, December 9 and 10, 2026, live online. Each day has two two-hour sessions: 9:30 to 11:30 a.m. and 1 to 3 p.m. Eastern. That is eight hours of instruction in all.
Every session is recorded. Transcript included. Recordings and transcripts stay available to paid registrants for at least six months, so you can catch up on anything you miss. Each topic is taught, demonstrated live, and then built by you. Optional lab sessions run for 30 minutes before each main session if you want extra build time.
Yes. The sessions are live with the instructor.
No. Heflin Group is independent, and not affiliated with, sponsored by, or endorsed by Anthropic. The course is taught and demonstrated using Claude Code, a product of Anthropic.
You need your own paid Claude plan from Anthropic, either Pro or Max, a Windows or Mac laptop where you can install programs, and a real task of your own, using sample material. You sign up with Anthropic, not with Heflin Group, and Anthropic bills you directly, separate from the $495; it sets the current price at claude.com/pricing. Claude Code comes with Pro and Max. A setup checklist arrives by email after you pay.
Either. Claude Code runs on macOS 13 or newer and on Windows 10 (version 1809) or newer, with 4 GB of RAM or more and an internet connection.
You need a laptop where you can install programs; the technical name is “admin rights.” Most personal laptops have them. If yours is managed by someone else, ask whoever manages it, or use a personal laptop.
Your seat is confirmed the moment payment completes. You then get a confirmation email with the meeting link, a setup checklist, and an invite to an optional pre-class tech-support session. Then you show up on December 9 and 10, ready to build.
Full details are in the Terms of Participation.
Join the waitlist and tell us the timeframe that would work. We'll let you know when the next cohort is scheduled.
Email customerSvc@heflingroup.net.
You leave with a working setup on your own laptop, templates to keep building, a way to back up what you built, and a simple plan for what to build next. Recordings and transcripts stay available to paid registrants for at least six months.
Yes. Heflin Group runs a Slack community for participants from every cohort. It's a place to ask questions and be inspired as you continue to build on your own.
Recordings and transcripts stay available to paid registrants for at least six months.
You need your own paid Claude plan (Pro or Max), billed by Anthropic and separate from this price.
Full details are in the Terms of Participation.
Dec 9–10 not going to work? Get on the list — tell us the timeframe that would, and we'll let you know when the next cohort is scheduled. We open new cohorts where the interest is.
Waitlist for Future Cohorts