Where should your team use AI next?


Hello Reader

Imagine hiring someone to redesign your kitchen. They arrive with a beautiful plan, a list of expensive appliances and a date to start work, but they haven’t asked how you use the space or what frustrates you about the kitchen you have.

Would you approve the plan?

As AI advisor Liza Adams says, “It makes more sense to reimagine the work first. A built-in pet-feeding station and an eight-burner range might look impressive, but why would you put them in a kitchen for a family that doesn’t have a dog and cooks a few times a year? Once you reimagine the work, then you can more easily reimagine the roles and org next.”

The AI version of that mistake is easy to make.

We see another team’s workflow and rush to copy it before asking whether it fits how our own team works.

If we focus only on how quickly someone can produce an output, we can miss the work it creates for everyone who has to review, correct and use it.

A Stanford Social Media Lab study found that 40% of U.S. desk workers said they’d received poor-quality AI-generated work in the previous month and they reported spending roughly two hours resolving each incident.

​Noah Brier, co-founder of Alephic, says we need to consider how much time we’re asking someone else to spend on our output.

“A meeting recap that takes no effort to generate and sits unread is fine, but if you spend five minutes generating a strategy document and your CEO needs 45 minutes to decipher it, you’ve created AI slop.”


Enjoying AI at Work? Share with a friend or colleague!


Start with the design

​McKinsey’s October 2026 AI value curve estimates a $2.52 trillion annual global opportunity from applying AI to existing work. Its calculation includes the cost of human review and rework, so the time spent checking and correcting an output counts against the savings.

Capturing that value requires decisions about which work to redesign, what a useful result costs and how your team will use the time saved.

McKinsey recommends five actions:

  1. Map the work at task level. Identify what consumes time, where human judgment matters and where AI could change cost or quality.
  2. Prioritize the strongest business cases. Consider the volume of work and savings per task.
  3. Measure cost per useful outcome. Include AI usage, human review and rework.
  4. Test changing costs. Check whether the business case still works if AI becomes more expensive.
  5. Decide where saved time will go. Identify the work people will take on and the skills they’ll need before deployment.

Bring together the people who create, review and receive the final outputs.

For lead follow-up, that might include demand generation, marketing operations and the sales rep receiving the lead.

Identify your value hypothesis

Which metric does your team need to improve?

For a demand generation team, that might be the percentage of qualified leads who book a sales meeting. For a content team, it might be the cost of producing an approved asset or the time between receiving a brief and publishing the work.

​Section calls this a value hypothesis.

It connects the work you plan to change with the result that justifies the investment.

Here’s how you might write out a value hypothesis:

Over six months, we’ll test whether AI-assisted lead research and follow-up can increase our MQL-to-meeting rate from 9% to 11%, within our current team capacity and qualification standards.

Your own hypothesis should use your team’s baseline, target and review date, then work backwards from the result.

Prioritizing workflow optimization with the three Vs

You’ll probably identify several places where AI could help. Section’s three Vs shows you where to start.

For our lead-follow-up example, account research might be a useful starting point if the team already has access to the sources and a clear standard for a good account brief.

A workflow that requires a CRM migration, new data contracts and several departments’ approval has a different implementation burden.

I’d choose an initial test with a meaningful business purpose, accessible inputs and an owner who has time to work on it.

Map the work before you rebuild it.

​Paige O’Neill, CMO of Culture Amp, mapped their marketing work into about 5,000 units and 36 scenarios. The exercise exposed duplication and breakdowns in collaboration, helping the team identify what needed to change and where AI could contribute.

Bring together the people who create, review and receive the final outputs. For lead follow-up, that might include demand generation, marketing operations and the sales rep receiving the lead.

Ask the team to pull up three recent leads, along with the CRM records, research and messages associated with each one. Reconstruct what happened from the moment each lead qualified through to the outcome, including every review and handoff.

For each step, record the input, owner, output and time involved. Separate working time from waiting time. Look for repeated delays, missing information and work that gets sent back for revisions.

Your map might uncover questions like these:

A five-minute writing task followed by a two-day handoff is a different problem than an hour spent gathering account information.

Use the map to remove unnecessary steps and define AI’s contribution. As an example:

  • You might have AI assemble a source-linked account brief, identify missing information and draft outreach using approved messaging.
  • The person’s responsibilities might include checking the evidence, deciding whether the approach fits and approving the message.

Test the design on real work so you can see what AI missed, what the reviewer corrected and what context had to be supplied.

Innovate with lighthouse teams

To get the most value out of AI adoption you need a two pronged approach:

  1. Drive broad employee adoption and capability
  2. Create lighthouse teams who redesign work and find innovation opportunities

Chief AI Enablement Officer at Section, Lauren Kaufman Witten says these approaches create value in different ways, but also reinforce each other.

“Broad capability makes new ways of working easier to adopt,” she says. “And lighthouse teams show what's possible through deeper redesign, and then generate lessons learned that other teams can apply.”

For each workflow, establish an owner, a target and a baseline before making changes.

The lighthouse team decides where AI can contribute and where people need to review, make decisions or handle exceptions. They test the redesigned process on real work and measure its effect on speed, quality, cost and business results.

As they learn, they document the instructions, tools, permissions and review requirements so other teams can build on what works.

Measure the result

When measuring the redesigned workflow you need to consider improvements in business metrics and additional team capacity.

If we go back to our earlier example, a marketing team aims to grow pipeline from $180 million to $360 million over 24 months. The plan assumes AI frees capacity equivalent to 50 full-time roles and the MQL-to-meeting rate improves from 9% to 11%. That would allow the team to reach its target with 175 people, 50 fewer than originally planned, avoiding $7.5 million in annual run-rate costs.

For your redesign pilot, track the chosen metric alongside the quality of the result. Where possible, compare similar groups over the same period and record other changes that affect performance.

​Katrina Wong, CMO of New Relic says to compare the cost of the existing process with the AI version.

“Make sure you count the labor, review and rework, software usage and upkeep,” she says. “You need the whole calculation to decide whether the change is worthwhile.”

Keep one-time setup separate from recurring effort and include the time spent preparing sources, testing instructions and training colleagues.

Where to start

Choose a business priority that warrants changing how your team works. Agree on what should improve over the next six months, then identify the core workflows that influence that result.

Give a lighthouse team responsibility for the redesign, along with the time and authority to change processes, handoffs and responsibilities. The people doing the work need a say in those decisions, training for their changing roles and time to test the new approach.

Decide how you’ll use any capacity you recover.

That might mean more customer research, faster follow-up or taking on work you’ve repeatedly postponed. Make those choices part of the plan so the team understands what the redesign is intended to achieve.

The takeaway: Redesign the work, the handoffs and the responsibilities together. Measure success in business results and the team’s capacity to deliver them.

More resources

For a deeper look at how this changes your marketing organization, read The AI-Ready Marketing Organization: A Strategy-First Playbook for B2B CMOs, by Drew Neisser and the CMO Huddles Future of Marketing Organizations task force.

It includes decision worksheets and a scorecard to help you assess your team’s readiness and plan changes to workflows, roles and responsibilities.

Poll

Kimber Spradlin, CMO at Graylog, wrote this after our Claude creative workshop:

"Extremely prescriptive. You walk away with a working system for building on-brand creative using Claude."

On Tuesday Nov 10 at 3pm PT, we'll set up Claude with memory and global instructions, build the Skills that keep copy and visuals on-brand and take one brief through to creative direction and images.

The workshop is rated 4.9 out of 5 on Maven.

Register even if you can't make it live and the recording comes to you.

Save my seat → Register now​

Did some one forward you this email? You can subscribe here.​

2120 Contra Costa Blvd #1059 , Pleasant Hill, CA 94523
​Unsubscribe · Preferences​

AI at Work

AI at Work is a weekly newsletter on how marketing teams redesign workflows, roles, and systems with AI. Real examples, practical frameworks, and repeatable processes operators can use immediately. Join thousands of successful marketing leaders by subscribing below!

Read more from AI at Work
AI and performance reviews

Hello Reader A quick note: Next week, you’ll notice a new sender email address: tahnee@perryco.ai. It’s still me and your weekly AI at Work. Add it to your contacts so you never miss an issue! ~ Performance reviews are coming up, and AI can make the preparation considerably easier. It can gather project records, summarize feedback and produce a polished draft in minutes. That’s appealing when reviews have long been criticized for taking too much time and doing too little to improve...

AI and employee revenue

Hello Reader When I was a SaaS marketing VP, our company was working toward $1 million in revenue per employee. If we weren’t hitting that number, then we weren’t approved for new headcount. Which meant I needed to push my team to be as productive as possible. Anthropic recently reported more than $6.6 million in revenue per employee for the last 12 months on revenue of $16.3 billion. I can imagine the conversation in leadership meetings now: “If Anthropic can do that, why can’t we get to $3...

Using AI

Hello Reader I teach AI for a living, and I still sometimes wonder whether I’m learning the right things. You might already use AI for research, campaign planning and content production. You’re getting better results. Then someone shares an AI agent that connects their customer data, runs research and produces a campaign plan, and suddenly your own work feels basic. Now you’re wondering whether to improve what you’ve built, learn a different tool or investigate something you hadn’t considered...