Agent Workforce · 4-week live course · No code

Learn to design, build, test and run an agent workforce.

The skill companies will need in 2027. Start with a working 20-agent org on day one, build your own department, and leave with evals you can show a client.

★★★★★ 4.6 513 ratings on Maven 2,500+ grads · 20+ cohorts · 15+ countries
Given day oneThe agent organization you inherit
20agents 94skills 5departments 1Chief of Staff 1dashboard to run it

AI Build Lab students have come from teams at

StarzCitibankUBSGoogleDocusignGeneral AtomicsOtsuka Precision HealthThe Glimpse GroupGeorgetown UniversityNTTVega Federal SolutionsRespageLuminalGroupLoopAchieveThirty & Co.SENSASIA SpasTapestry PeopleEH SolutionsSweet Spot Business AdvisoryASAI ServicesFuture CoachCTRN.eco

Why this skill, why now

Three things are true going into 2027.

01

Agents are exploding.

  • 15 → 150,000agents in use at the average Fortune 500 company, 2025 to 2028Gartner prediction, Apr 2026
  • 17% → 37%CIOs with AI agents deployed, in one year. Another 34% plan to within 12 months.Gartner CIO surveys, n=2,501 and n=2,002
  • $16.5B → $65.5Bspending on AI agents and assistants, 2025 to 2027Gartner forecast, Sept 2026

02

Agent orgs are inevitable.

  • 96%of organizations already run AI agents. Only 12% have one place to manage them all.OutSystems 2026, ~1,900 IT leaders (vendor survey)
  • +102%better results when agents were organized like a company instead of a flat group, with 74% fewer tokensOrgAgent, CUHK + IBM Research, 2026
  • 79%of multi-agent failures came from unclear roles and poor coordination. Not the AI model.UC Berkeley, 1,600+ failed runs

03

The people who build them are indispensable.

  • 7%of companies let agents run with nobody in the loop. Everyone else needs someone who can direct them.Bain, 951 companies, 2026
  • Org designWhen groups of agents fail, it's usually an org design problem: who does what, how work is handed off, who checks it. That's a skill. It's the one this course teaches.

The skills

Ten skills behind every agent org that works.

The first nine you learn hands-on, on your own work. The tenth, taking it to a client, comes in the bonus Implementation Track.

Design01

Process mapping

Break a real job into steps. Decide which go to an agent and which stay with a person.

Week 2
Design02

Org design

Roles, departments, a Chief of Staff who routes the work, and how work passes between agents.

Weeks 1 & 3
Design03

Context engineering

What each agent needs to know, in one shared library, kept current and traced to the source.

Weeks 1–2
Design04

Agent instructions & role cards

What each agent does, what it knows, what it hands back. Plus reusable skills (SKILL.md).

Week 3
Control05

Guardrails

What needs a person, how much each agent does alone, who may touch what.

Weeks 1 & 3
Control06

Tools & connectors

The right tool to the right agent: Gmail, Outlook, your files, with access kept safe.

Weeks 2–3
Proof07

Evals

Golden examples, observational evals and pass/fail checks, so you know it works every time.

Weeks 3–4
Proof08

Tracing & improvement loops

Read every run in Langfuse, find what broke, fix it, run it again.

Weeks 2 & 4
Run09

Agent operations

Scheduled and overnight runs, a daily health check, and a dashboard.

Weeks 3–4
Sell10

Discovery, scoping & ROI

Run client intake, scope the project, prove the ROI and price the rollout.

Implementation Track · soon

What you can put on your resume

“Designed and deployed a 20-agent AI workforce in Claude Code. Built custom agent skills, set up Langfuse tracing, and ran LLM-as-judge evaluations with golden sets and regression tests.”

Tools you'll use Claude Code · Codex · GitHub · Langfuse · Gmail / Outlook connectors

From our September 2026 Agent Workforce cohort

“Challenging AND absolutely doable.”

E
EmilyAI Systems Strategist · EH Solutions
Maven review
★★★★★Live cohort · Sep 2026
So many, including myself, do not have a developer or engineer background. Yes, it is challenging AND absolutely doable!! What I am building and doing with AI after 4 weeks, with safety, human judgement, and documentation baked into each process, still blows my mind!
A
AgusSoftware Engineer · Big Data Pricing
Maven review
★★★★★Live cohort · Sep 2026
I love that it's designed for people starting from zero experience. You don't need any background to get real value out of it. The team was incredibly kind and helpful, and they really know their stuff.
J
JillAI Marketing Transformation · Own business
Maven review
★★★★★Live cohort · Sep 2026
Just wrapped up my third course with Sara and Tyler and each one just keeps getting better! If you want to learn how to build AI agents that actually do real work, this course is for you.

What you walk out with

A complete agent organization. Yours to run or sell.

Day 1

A distilled version of our org, made yours

Departments mapped, seats filled, a Chief of Staff routing by your rules. Installed on day one.

Week 1

A Chief of Staff in your voice

One seat you talk to. It routes the work and holds what needs your yes.

Week 2

An org that knows your business

Your own material connected and organized, and an SOP your agents follow.

Week 3

Three agent teams on real work

An email team on your inbox, a department you design and build yourself, and a team built overnight.

Week 4

Your dashboard

One screen to see the org running, read the receipts and act on what needs you.

Week 4

Proof it works

Eval scores, traces, receipts and a proof page. Not a demo that worked once.

Forever

The method

The part that outlasts the four weeks: you can keep adding departments on your own.

Forever

Yours to run or sell

Run it for your business, or install it for clients. Models change. Your org stays.

The big idea · your dashboard

Run your whole org from one screen.

You don't babysit chat windows. You open your dashboard, see what moved overnight, and make the one decision that needs you.

  • A morning brief: what your agents did, and what needs your yes
  • A queue of decisions, each with the receipt attached
  • Your org, your knowledge, your projects and your evals in one place
  • Message your Chief of Staff from anywhere on the page
Week 4 · Open your dashboard
agent workforce · dashboard
The Agent Workforce dashboard: a CEO overview saying good morning, one decision in your queue, the item that needs you now with its state and your move, and company momentum below.

The full curriculum · as it appears in the course

4 weeks. 21 modules. 51 lessons.

This is the course portal, week by week. Every lesson comes with a ready-to-paste start prompt, and The Professor walks you through it. Expect 4–6 hours a week, plus live sessions.

W1

Meet and guide your organization

4 modules · 8 lessons

Complete the orientation, then meet the organization: twenty agents, each met the first time it works. Across eight lessons you use approved context, hire and hear your Chief of Staff, adopt a full charter and compare actual work.

15 stepsLive session weeklyHomework huddle

You leave withAn organization that takes work from you, a Chief of Staff hired for your work, and a charter every agent follows.

01
Core CurriculumThe course. Do these first, in order.
Required

Your week, by the clockNine stops, about 5½ hours of lessons

Each card is as wide as its lesson is long. Spread it across your week. Orange marks the one long stretch.

  1. 9%OrientationMeet your organization~30 min
  2. 11%Lesson 1First job~35 min
  3. 15%Lesson 2Reusable instructions~50 min · longest
  4. 9%Lesson 3A place for everything~30 min
  5. 6%Lesson 4A fact, a question~20 min
  6. 8%Lesson 5Your Chief of Staff~25 min
  7. 14%Lesson 6The hire~45 min
  8. 14%Lesson 7The charter~45 min
  9. 14%Lesson 8The chart moves~45 min
Module 0Before you startRead the welcome, then get your organization into your workbench.
1Before you startLesson
2Welcome to Agent WorkforceLesson
3Get your organizationLesson
Module 1Meet your organizationTour the organization, try a small job and decide what needs attention.
1OrientationOpen the organization for the first time: the pieces, the chart, the tour, the before photo.Lesson
2Give your organization a first jobTry a small job, check what comes back, and decide what happens next.Lesson
3Put reusable instructions in the right placeMove reusable instructions into the right home, and keep an honest starting line before making changes.Lesson
Module 2Tell your organization about your workGive information a useful place, describe your work, and check one answer against a fact you supplied.
1Give everything a placeDecide who needs each piece of information, then give the organization a brief it can use.Lesson
2Give it a fact, ask it a questionCheck one answer against a fact you supplied and decide where a correction belongs.Lesson
Module 3Make your Chief of Staff work for youUnderstand why the Chief of Staff sits between you and the organization, then hire it for your work.
1Understand your Chief of StaffExplain what your Chief of Staff is for and follow one request through the choices it makes.Lesson
2Hire your Chief of Staff and choose its styleTell it what matters to you this month, watch it plan the same request again, and choose how its decisions reach you.Lesson
02
BonusSafe to skip. Nothing later in the course needs it.
Optional
+Voice typing: talk to your AI anywhere8 min
+Your Professor: change how it explains things5 min
+Your data: where every copy goes, and how to clean it6 min
+GitHub masterclass12 min

Look inside the course

You don't watch videos about agents. You run them.

Where the value moved

Building is the first step. The value is at the top of the stairs.

  1. Step 1BuildInstructions and tools for one agent. Tools now help a lot.
  2. Step 2DesignWhich job goes to which agent, and how work passes between them.
  3. Step 3TestProof it works every time, not once.
  4. Step 4GovernWho may do what, what needs a person, what gets logged.
  5. Step 5Evaluate & improveScore the work, fix what drifts, rerun, add the next agent. Paid every month.

Most courses stop at step 1. This one takes you all five.

What you do with it

Two ways in. Both doors are open.

Path 1 · Sell it

Build agent orgs for clients.

Pick a kind of business you already know. Build their agent organization, prove it works, then run it for them.

You don't sell hours, and you don't sell agents. You sell a measured result and the running of it.

Path 2 · Get hired for it

Run it inside a company.

Most companies grow this skill inside, in people they already employ. 47% of workers already spend more time directing AI than doing the work themselves. BCG, 11,749 workers

Doing it well, with proof, sets you apart.

Who's charging what

People are already selling this.

From their own words and price pages.

ModelWhoAsking price
Managed, monthlyOrgo (Nick Vasilescu)$5K/mo
Managed, up to 5 agentsLeadByAI · price page$3,750 + $5K/mo
Build + run, company of 5–15Affixed AI · price page$75K + $7.5K/mo
Build + run, 140-person firmGigabit · case on price page$156K + $4.5K/mo
Sprint + embedded, $5M–$50M firmsAwarded · price page$15K–40K + $8K–15K/mo

Their published prices and stated models, read Sept to Oct 2026. Asking prices and their claims, not audited results or what anyone will earn.

Your proof decides your price. Agent Workforce gives you the proof: evals, traces and a proof page you can show a client. The bonus Implementation Track, coming soon, walks you through a full client engagement: intake, scoping, ROI and pricing.

You won't do this alone

Live, every week. With people who've done it.

LIVE

Live sessions with Sara and Tyler

Instruction and Q&A every week, plus office hours. Everything is recorded, so you never miss a thing.

HW

Homework huddles

Drop-in clinics run by our team. Bring your project, your error, your question. Leave unstuck.

2K+

A community of 2,000+ peers

Our alumni Slack is where people share builds, hire each other and keep going long after the course ends.

You stay in control

Agents do the work. You make the calls.

Safety isn't an add-on in this course. It's how the organization is built.

Approval gatesAgents draft and propose. You approve, change or say no.
Nothing surprising overnightThe overnight build needs your yes to start, and your yes again before anything is installed.
You send your own emailYour email team drafts. You review and send.
Receipts for every runSee exactly what your team did, what it touched and why.

Is this technical? How it actually works

No code. A folder, an app, and you.

We run our own business agent-native

Five people. The output of a 60-person company.

5people on the AI Build Lab team
9,700+hours of work produced by our agent workforce
~60people it would take a traditional company

That's the ROI of running agent-native. It's the same way of working this course teaches you to build.

Your instructors

Meet Sara and Tyler. Practitioners, not theorists.

We build agent systems for our own business and for clients, and we teach what we use.

Sara Davison

Sara Davison

Co-founder · Strategy lead

Sara helps non-technical professionals become agentic AI practitioners, through frameworks and hands-on execution, not theory. She designs the curriculum and turns complex AI concepts into mental models you can actually use.

Tyler Fisk

Tyler Fisk

Co-founder · Lead instructor

Tyler spent 20+ years turning startups into scale machines before becoming a practitioner in agentic AI systems. He teaches by doing: live demos, real-time debugging, mistakes and all. He'll roast your agent architecture, then help you fix it.

And the rest of the five-person team behind your cohort

Hunter
HunterThe build, the course and the demo
Wade Murley
Wade MurleyYour home base, Essentials and delivery
Michael
MichaelInfrastructure, repos and the installer
2,500+Enrollments
4.6★513 Maven ratings
2,000+Community members
25,000+People taught in our workshops

What our students say

Senior people. Busy people. Non-technical people.

DO
Director of Product ManagementStarz · AI Build Lab grad
Student review
★★★★★AI Build Lab grad
Hands down the best course I've ever taken (for context, I've taken a wide variety from Product School, MIT exPro, Nielsen Norman, etc). Sara and Tyler are extremely knowledgeable, open hearted, and fun.
BD
Business DirectorSENSASIA Spas · AI Build Lab grad
Student review
★★★★★AI Build Lab grad
I was terrified going into this course that I was trying to learn something way beyond my technical capability, but I was incredibly wrong! I've built our company an actual customer service agent.
A
AnneFounder & CEO · CTRN.eco
Maven review
★★★★★Live cohort · Sep 2026
The real five-star experience was the people! The instructors bring an extraordinary level of passion, enthusiasm, and genuine commitment to AI. They truly live and breathe this work.
C
COOCitibank · AI Build Lab grad
Student review
★★★★★AI Build Lab grad
This is the real deal if you look to build skills that you can actually use at work.
F
FounderProfitable Popularity · AI Build Lab grad
Student review
★★★★★AI Build Lab grad
I just completed my first agentic workflow coming from someone who had never built anything in AI as of 3 weeks ago.
VS
VP Strategy and OpsThirty & Co. · AI Build Lab grad
Student review
★★★★★AI Build Lab grad
The best value for any bootcamp, workshop, and learning module I've ever paid for.
C
ChristineHR Consultant · Tapestry People
Maven review
★★★★★Live cohort · Sep 2026
AIBL's courses are always amazing. They are generous with their knowledge, time and team resources.
DA
Director, AI StrategyASAI Services · AI Build Lab grad
Student review
★★★★★AI Build Lab grad
This is the seventh course I've invested in on Maven, and it's by far one of the best. I'd give it an 11 out of 10!
PC
PhD Candidate, AI in HRAI Build Lab grad
Student review
★★★★★AI Build Lab grad
I've accumulated over 10 certifications in AI. This course is the best of these courses hands down.

Is this for you?

Built for people who run things, not people who write code.

Consultants, agencies and implementers

You want to build agent organizations for clients as a high-value service. What you build is yours to deploy and resell.

Professionals who want the role

You want to be the person your company trusts to design, run and govern its agents, with proof of what you can do.

Founders and operators

You want an agent org running the day-to-day in your own business, and the know-how to build the next one.

This isn't for you if…

  • You want a prompt pack or a one-hour tutorial. This is four weeks of building.
  • You want to watch, not do. You'll build on your own work every week.
  • You can't set aside 4–6 hours a week for the next month.

Enroll

Everything you need to run an AI workforce.

What's included

  • Your starter agent organization20 agents, a Chief of Staff and a charter, wired and ready on day one. Yours forever.
  • 4 weeks · 21 modules · 51 lessonsStep by step, with copy-and-paste start prompts, in Claude Code or Codex.
  • Live sessions with Sara and TylerWeekly instruction, Q&A and office hours. All recorded.
  • Homework huddlesDrop-in clinics for help on your project.
  • The ProfessorYour course guide agent, in every lesson.
  • Agent Native EssentialsOur onboarding course takes you from zero before the cohort starts.
  • Bonus lessonsVoice typing, data hygiene, GitHub masterclass, campaign walkthrough and more.
  • Lifetime accessCome back to the lessons and recordings whenever you need them.
  • Alumni Slack community2,000+ peers who share, collaborate and hire each other.
  • One more cohort, freeTake the course again with a later cohort to embed what you learned.
  • Certificate of completionShare it with your employer or on LinkedIn.

Agent Workforce

$1,950

Enrollment is on Maven, where you choose your cohort. Ask your employer: Maven provides an expense template, receipts and a certificate of completion.

Maven
guarantee
Backed by the Maven Guarantee.Your purchase is covered by Maven's satisfaction guarantee.

Secure checkout on Maven

Questions

Before you decide.

I'm not technical. Can I really do this?

Yes. That's who this is built for. You never write code. Every lesson gives you a prompt to paste, The Professor walks you through it, and the live sessions and homework huddles catch you if you get stuck. Our students are operators, founders, consultants and executives, not developers.

How much time does it take?

Plan on 4–6 hours a week for lessons and your project, plus the live sessions. Your first week is about 5½ hours of lessons. It's immersive, and it's designed to fit around a full-time job.

What if I can't make a live session?

Everything is recorded and added to the course portal, so you can catch up any time.

What do I need before I start?

A computer, a paid Claude or ChatGPT plan (so you can use Claude Code or Codex), and a free GitHub account. Agent Native Essentials, our free onboarding course, is included and sets up your workspace before the cohort begins.

Is this just chatbots with job titles?

No. Your agents hand work to each other, follow a shared charter, use real tools and get checked by a reviewer. In Week 4 you prove the work is good with grades and tests, not a gut feeling.

Can my agents do things without my approval?

Not unless you say so. Agents draft and propose; you approve. Your email team drafts and you send. Anything built while you're away waits for your yes before it's installed.

Where does my data live?

Your organization lives in your own workspace and your own GitHub account. You choose what it can read. A bonus lesson shows exactly where every copy of your work goes and how to clean it up.

Can I use this with clients?

Yes. What you build is yours to deploy and resell, and the method works for any business.

Does this course teach me how to implement this for clients?

Yes. A bonus Implementation Track is coming soon. After Week 4, you take everything you built for yourself and run a full engagement for a fictional company, from the first email to the handover: client intake and discovery, scoping, designing and standing up their agent organization, proving the ROI with a pilot, pricing the rollout and presenting the business case. You finish knowing you can do it again for a paying client.

When does it start?

Cohorts run throughout the year. See the current dates and choose your cohort on the Maven course page. Your enrollment includes one more cohort, free, so you can take it again later.

What if it's not right for me?

Your purchase is backed by the Maven Guarantee.

How is this different from Agent Native Camp?

Camp is depth: you build your own always-on fleet, on your own setup, and go deep on one project. Agent Workforce is breadth: you get our working agent organization in Claude Code or Codex, with no servers to run, and learn the method to add team after team. Build for yourself in Camp. Build something you can hand over in Agent Workforce.

Learn to build it

The people who build agent orgs
are indispensable. Be one.

Four weeks from now you could have a working agent organization, your own department built from scratch, and the proof that it delivers. The skill companies will need in 2027.

Maven Guarantee · One more cohort free · No code

The agent organization you inherit, full size.