Good morning,
The Week the Agent Era Stopped Being a Forecast…
Two stories landed this week that change the conversation. On Tuesday, Google opened I/O 2026 by unveiling Gemini 3.5 Flash and Antigravity, its agent-first development platform, alongside information agents in Search - personal AI agents that run in the background, monitor topics, and take action on your behalf. Sundar Pichai framed the shift in one line: we are moving from AI that helps you write to agents that help you act.
Two days later, Nvidia reported $81.6 billion in quarterly revenue, up 85% year over year. On the earnings call, Jensen Huang told the Street that agentic AI has arrived, doing productive work, generating real value and scaling rapidly across companies and industries. He closed with a sharper line: demand has gone parabolic - the reason is simple, agentic AI has arrived. Nvidia also introduced Vera, a CPU it built specifically for agentic workloads.
What it means for your team. The two largest players in software and silicon just told the market the same thing in the same week. The shift from tools to agents is no longer the next thing - it is the current thing. The companies hiring agents now are setting the operating model your competitors will run against you 12 months from now. The rest of this workshop is how you build yours.
FOR SELLERS Stop using AI to draft emails faster. Start using it to research accounts, prep calls, and follow up while you sleep. Move from prompt-and-paste to agent-and-review. |
FOR MANAGERS Your reps are using AI. You just don't know how. Build a simple inventory of where agents are running on your team, what they produce, and which ones earn their keep. |
FOR OWNERS Agents change cost structure. A small team with the right agentic stack can outproduce a large team without one. The question is no longer headcount. It's leverage. |
The revenue impact is direct. Sellers using agents to handle research, prep, and follow-up reclaim hours per week. Managers running agentic pipeline reviews catch deals earlier. Owners who build the right stack lower cost-per-close. The teams pulling away from the pack treat agents as staff, not software.
Key Terms
Tool | Software that does one thing when you ask. ChatGPT writing an email is a tool. You drive every step. |
Agent | Software that takes a goal and runs a sequence of steps on your behalf. It plans, executes, and reports back. You set direction; it handles execution. |
Agentic Stack | The set of agents a person or team runs together. Each agent owns a defined job. The stack functions like a small operating team. |
Do's and Don'ts
Do | Don't |
|---|---|
Give each agent one clear job with a defined output. | Build one giant agent that tries to do everything at once. |
Review what every agent produces before it goes out the door. | Set agents running and trust them blindly the first week. |
This Week's Challenge
Pick one task you do every week that follows a pattern. Build one agent to handle it. Run it five times before next Sunday.
The Workshop
MODULE 01 · FOR SELLERS
Hire Your First Agent
Pick one task. Give it to an agent. Win back your week. You start every Monday with a list of accounts to research, calls to prep, and follow-ups you owe. By Wednesday, half of that list is still untouched. The work is not hard. There is just too much of it. An agent built for one of those jobs runs in the background while you sell. Which task on your list this week could a teammate run for you if you had one?
EXAMPLE A digital ad seller at a regional publisher spent four hours every Monday researching her top ten accounts. She built a simple research agent: ticker, recent news, last quarter results, key hires, ad spend signals. She now starts Monday with ten briefs already done. She uses those four hours to make calls. |
The Steps
Step 1. Define the job before you build the agent.
Pick one task that eats your time and follows a pattern. Research. Recap. Follow-up draft. Write down the inputs, the steps, and the output. If you cannot describe the job in three lines, you are not ready to build the agent. Clarity on the job is 80% of the work.
Step 2. Build it, review it, and use it five times.
The seller above did not buy enterprise software. She built her account research agent in a free tool over a lunch hour. She used it once, fixed the output, used it again, fixed it again. By the fifth run, the output was better than what she used to produce herself. She now runs that agent every Monday morning before she opens her inbox.
DO THIS TODAY Pick the one task from this week you wish someone else had handled. Write it down in three lines: input, steps, output. That is your first agent's job description. SHARE IT Forward this newsletter to one peer on your team and tell them what task you picked. They will tell you theirs. Compare notes on Friday. |
Q. What if my company will not let me use AI tools?
A. Find out what is approved and use that. Most companies have at least one sanctioned tool. The point is not the tool. The point is the discipline of defining a job and handing it off. That habit is what changes your week.
MODULE 02 · FOR MANAGERS
Inventory The Stack
Find every agent already running on your team. Decide which ones earn their keep. Your reps are already using AI. Some of them have been at it for months. You have no idea what they have built, what it produces, or what it pulls from your CRM. That is not delegation. That is drift. Your job is to surface it, sort it, and turn it into a team asset. If your top rep walked out tomorrow, would the agents they built leave with them?
EXAMPLE A sales manager at a B2B publisher ran a 30-minute audit with each rep. She asked one question: show me every AI prompt or workflow you use weekly. Eleven reps surfaced 47 different agents. Three were genuinely useful. The rest were one-time experiments. She kept the three, documented them, and shared them across the team. Average prep time per call dropped by 40%. |
The Steps
Step 1. Run the audit before you build the policy.
Skip the policy memo. Spend 20 minutes per rep asking what they actually use. No judgment, no compliance frame. You are taking inventory of a shadow operation that is already shipping work. Most managers find a mix of useful agents, abandoned experiments, and a few risks. You cannot manage what you cannot see.
Step 2. Promote the winners. Retire the rest.
The manager above did not standardize on enterprise tools. She picked the three agents that worked, wrote one-paragraph documentation for each, and shared them in a shared doc. New hires now start with three working agents on day one. The reps who built them got credit. The agents that did not earn their keep got retired.
DO THIS TODAY Schedule one 20-minute meeting with your top rep this week. Ask the one question above. Bring no agenda. Just listen and take notes. SHARE IT Send this to one other sales manager. Tell them what you found in your top rep's stack. Their reaction will tell you whether your team is ahead or behind. |
Q. What if my reps are using AI in ways that violate compliance?
A. Better that you find out now than in an audit. Use the inventory as the starting point for a policy. Lead with curiosity, not punishment. Reps hide what they think will get them in trouble. They share what they think will make them look smart.
MODULE 03 · FOR OWNERS
Build For Leverage
Treat agents as staff. Build the operating model that pulls them into the work. You are looking at your headcount plan for next year. You see what every role costs and what it produces. There is a third column you have not added yet: which roles could a small team plus an agentic stack outproduce? The teams pulling ahead in your category are already running that math. If you had to grow revenue 30% with the headcount you have today, where would the agents go first?
EXAMPLE A media company CEO froze hiring on his SDR team for two quarters. Instead, he gave his two existing SDRs a budget for tools and a mandate: build the agentic stack that lets the two of you outproduce a team of six. They built agents for account research, list building, and personalized first-touch outreach. Pipeline generated per SDR doubled. The CEO redeployed the two open headcount to closing roles. |
The Steps
Step 1. Pick the function. Then design the operating model.
Do not sprinkle agents across the whole org. Pick one function where output is measurable and the work is repetitive. Research. Prospecting. Onboarding. Renewal prep. Design the operating model first: what humans do, what agents do, what gets reviewed and by whom. The model is the moat, not the agent.
Step 2. Fund the experiment and measure the output.
The CEO above did not buy a platform. He gave his SDRs eight weeks, a small tools budget, and a clear target: double pipeline per rep. He reviewed output weekly. The agents that drove the result got promoted into the standard SDR playbook. The ones that did not got cut. Eight weeks beats eight months of pilot.
DO THIS TODAY Pick one function on your org chart where output is measurable and work is repetitive. Write the question above on a sticky note and put it on your monitor. That is your starting point. SHARE IT Send this to one other owner or GM you trust. Ask them which function they would pick. Their answer will sharpen yours. |
Q. Won't this just create a different kind of overhead?
A. Only if you run it like a project instead of a function. Agents become overhead when no one owns the output. They become leverage when someone is on the hook for what they produce. Same rule that applies to every other hire on your team.
Recommended Reading
Co-Intelligence: Living and Working with AI Ethan Mollick · 2024 · Portfolio The playbook for treating AI as a coworker, not a tool. Mollick gives you the mental model that makes everything else in this newsletter make sense. |
The AI-Driven Leader Geoff Woods · 2024 · AI Thought Leadership How to use AI as a strategic thought partner. Built for owners, GMs, and senior leaders who need to lead the AI shift, not just authorize it. |
Revenue vs. Sales Mort Greenberg · 2024 The operator's framework for building a revenue org that scales. Where the agentic stack fits inside the broader system that produces revenue. |
==
The Revenue Workshop isn’t theory. It’s a field-tested system used by real leaders, in real markets, under real pressure.
Each newsletter is based on one of over 300 workshops and worksheets found in the eight books of the RevenueVsSales.com and TheFocusedSeller.com book series.
We’d love your help growing the list. Please share the newsletter with one person you think would benefit from it and ask them to sign up. Thank you!
To suggest workshops you’d like to read next, email [email protected]
Thank you!