Most teams are stuck in Stage 1 of AI adoption, where individuals use AI to speed up their own tasks but team workflows stay the same. Stage 2 happens when managers redesign processes end-to-end with AI embedded in them. Start with one recurring meeting, set shared team standards, and delegate to workflows, not just people. That’s the shift, and it’s a leadership job.
Key Takeaways
- Stage 1 (individual AI use) is necessary but not sufficient. The real productivity gains live in Stage 2 (workflow-level integration).
- Meetings are the easiest place to start. The gap between Stage 1 and Stage 2 is most visible there, and the redesign is contained.
- Without shared team standards, AI use stays fragmented and quality stays inconsistent across people.
- Delegate to workflows, not just people. Ask “what workflow handles this?” before assigning a task to a person.
- Speed without review creates risk. Decision ownership stays with humans, even when AI surfaces answers fast.
What’s the difference between Stage 1 and Stage 2 AI adoption?
Stage 1 is when individuals on your team use AI to speed up their own tasks. One person drafts emails with it. Another runs meeting notes through it. A third uses it to generate slide decks. Each person has their own tool, their own workflow, their own approach. The underlying process never changes.
Stage 2 is when AI is embedded in how the team operates, not just how individuals work. The output of one AI-assisted step feeds directly into the next one. The question shifts from “how can I use AI?” to “given that AI is available to all of us, how should this process actually work?”
If your team’s workflows haven’t changed since you introduced AI tools, you’re in Stage 1. It doesn’t matter how many people are using those tools individually.
Most teams adopted AI the same way organizations adopted calculators in the 1970s. Everyone got one, individual tasks got a little faster, and then nothing else changed. The processes stayed the same, the org structure stayed the same, and the calculator just sat on the desk doing one thing. That’s where most teams are right now.
Why are meetings the easiest place to start redesigning workflows?
Meetings are where the gap between Stage 1 and Stage 2 is most visible, and where the fix is most accessible. A typical team meeting looks like this: someone creates an agenda, maybe. People show up without reading the pre-read. Someone takes notes while also trying to participate. Action items get agreed on but captured inconsistently. Follow-ups land in different inboxes and get interpreted differently by different people.
That process can look completely different in Stage 2. AI generates the agenda based on standing context. It pre-reads the materials and creates a preparation document so people show up ready to make decisions rather than catch up. During the meeting, decisions and action items get logged in real time into whatever tool the team already uses, whether that’s Notion, Asana, or something else. After the meeting, the summary goes out automatically, action items are already assigned, and follow-up materials are drafted based on what was discussed.
The meeting itself doesn’t have to change much. What changes is everything around it.
Checkpoint: If someone on your team is using an AI recording tool but it’s not connected to anything else the team uses, that’s a calculator, not a workflow. Redesign it end-to-end or don’t count it as progress.
How do managers set shared standards for AI use across a team?
The most common mistake at this stage is letting everyone experiment independently with no shared standards. One person uses one tool. Another uses a different one. Output quality varies. Nobody shares what’s working because it feels like an individual edge.
Your job as the manager is to set guardrails. Not rigid rules. Shared standards. That means deciding as a team:
- What tools get used for what purposes
- What a status update or client deliverable looks like
- Where decisions get logged
- What the review process is before AI-generated content goes to clients or leadership
Transparency about AI use within the team is not optional. It’s a functional requirement. If you’re reviewing a report and you don’t know what was AI-generated and what wasn’t, or what kind of review it went through, you can’t do your job well.
Checkpoint: If a new hire would have to figure out each person’s individual AI workflow by asking around, you don’t have shared standards yet.
What does it mean to delegate to a workflow instead of a person?
The traditional delegation model is straightforward. Assign a task to a person and let them run with it. That model is becoming insufficient. The better question is what workflow should handle this, and who is responsible for that workflow.
Here’s a concrete example. A client meeting just happened. Follow-up materials need to go out. The Stage 1 approach is to hand it to someone who pastes the transcript into ChatGPT. The Stage 2 approach is different.
In Stage 2, the transcript connects to your CRM. The CRM pulls in client history, past conversations, and engagement data. AI generates a follow-up document that includes the context the client actually cares about. That document goes through a review stage before it reaches the client. The whole thing happens because a leader thought through the process and set it up deliberately. Not because one ambitious person decided to try something new.
When a new request comes in, ask whether it fits into an existing workflow before assigning it to a person. If no workflow exists for it, that’s the gap to address.
How do teams keep good judgment when AI surfaces answers fast?
Speed creates its own risk. When AI surfaces recommendations quickly, teams can start acting on them without enough critical review. That’s a cultural problem as much as an operational one.
Leaders need to be explicit about three things: which decisions can be made at the team level, which require escalation, and what gets reviewed before anyone acts.
Institutional context matters here. Newer team members and AI both lack the judgment that comes from years of doing the work. If something looks off to you based on experience, that instinct is worth slowing down for. The goal is a team that knows when to trust AI output and when to question it. That standard has to be set explicitly.
If you don’t set it, the team defaults to one of two bad outcomes: over-relying on AI without thinking critically, or second-guessing everything and not using it at all.
Checkpoint: If a decision is being justified primarily by what an AI recommended, that’s a signal your review process isn’t working yet.
What does the research say about workflow-level AI integration?
Three findings are worth keeping in your back pocket.
McKinsey Global Institute estimates that generative AI and existing technologies could automate work activities that absorb 60 to 70 percent of employees’ time today. The constraint on adoption is not missing technology. It’s organizational behavior. Most companies are still using AI as a task accelerator, not as a workflow redesign tool.
A 2026 working paper from MIT Sloan researchers makes a related point. Their research argues that AI’s biggest impact comes from how it reshapes entire workflows, specifically how tasks are sequenced, grouped, and handed off between humans and machines. Co-author Peyman Shahidi puts it plainly: “It’s not about how I’m going to introduce AI in my existing workflow. It’s about how I can redesign my workflow in such a way that is more AI-friendly.”
Separately, a 2026 report covered by MIT Sloan Management Review found that fewer than 3 percent of workers qualify as “AI practitioners” or “experts,” meaning employees who integrate AI into day-to-day workflows and generate meaningful productivity gains. Most of the rest are using AI for narrow tasks like drafting text or replacing search. The gap between potential and reality is almost entirely a workflow design problem.
The teams pulling ahead aren’t using better tools. They’re using the same tools with better thinking about the process.
What to do this week
Block two hours with your team. Map every recurring process, standard output, and regular task. For each one, ask whether you’d build it the same way if you were starting from scratch today with current AI tools available.
Pick one recurring meeting and redesign it end-to-end: agenda, preparation, in-meeting capture, follow-up. Run the new version for four weeks. Treat it as your proof of concept before expanding.
Document your team’s shared standards for AI use. Cover which tools get used for what, what outputs look like, where decisions get logged, and what the review process is before anything goes external.
The teams that will be well-positioned in the next few years aren’t the ones with better access to tools. They’re the ones where someone stepped back and asked how the work should actually be structured given what’s available. That’s a leadership job. It belongs to you, not to IT and not to each individual on your team.
FAQ
Do I have to standardize tools across my team to get to Stage 2?
You don’t have to use the exact same tools, but you do have to agree on what each tool is used for and what the output looks like. The problem is not tool diversity. The problem is inconsistency in how AI gets applied to the same kind of work.
How do I know if my team is in Stage 1 or Stage 2?
Look at one recurring process and ask whether the output of one AI-assisted step automatically feeds into the next one. If a person has to copy, paste, and re-prompt at every handoff, you’re in Stage 1. If the process flows end-to-end with AI embedded, you’re in Stage 2.
What’s the first workflow I should redesign?
Pick the one with the highest frequency and the most pain. For most teams, that’s a recurring meeting plus its follow-up. High frequency means you’ll iterate quickly. High pain means the team will notice the improvement.
How do I prevent my team from over-relying on AI for decisions?
State explicitly which decisions can be made at the team level, which require escalation, and what gets reviewed before action. Then ask your team to surface their reasoning, not just the AI’s recommendation, when they bring something to you.
What if my company doesn’t have an AI policy yet?
Set team-level standards anyway. You don’t need a company-wide policy to decide how your team uses AI for status updates, client deliverables, or meeting notes. Start with what you control.
Should I be driving this, or should it come from senior leadership?
Both. Senior leaders should set the broad guardrails. But the process redesign happens at the team level, because that’s where the workflows actually live. Waiting for top-down direction is one of the main reasons teams stay in Stage 1.
To help you run the conversation with your team, we built a Workflow Audit Guide that walks through the two-hour mapping exercise, the meeting redesign template, and the shared-standards checklist. Access it here.
