You bought the subscription six months ago.
You ran the training. You sent the launch email. You added it to the onboarding pack. You even put it in someone's performance review template.
And somehow, when you look at the usage data, only a handful of people are actually using it. The rest are doing things the same way they were a year ago, occasionally remembering the tool exists when they want to look productive in a meeting.
If this sounds familiar, you are sitting on one of the most expensive problems in modern operations. According to a recent Gallup poll, only eight percent of US workers use AI daily. Asana's Work Innovation Lab found that just thirty-six percent of employees use AI on a weekly basis. The Economist reported in late 2025 that the employment-weighted share of Americans using AI at work actually fell by a percentage point, with adoption dropping most sharply at companies with over two hundred and fifty employees.
This is not a technology problem. The tools work. It is an adoption problem, and the businesses that crack it pull dramatically ahead of the ones that do not.
This post is the tactical companion to broader AI change management. Where change management is about leading the strategic shift, this post is about the specific moves that turn "we have the tools" into "the team is actually using the tools." If you have already done the strategic work and the adoption is still stalling, the answers are here.
Why teams do not use the AI tools you bought
Before the tactics, it is worth being precise about what is actually going wrong when adoption stalls. The causes are surprisingly consistent.
The tool adds steps instead of removing them. Your team's current workflow takes twenty minutes. Using the new AI tool takes twenty-two minutes, because the team has to context-switch into the tool, prompt it, review the output, and integrate it back. The tool is slower than the existing process, even if it is theoretically better. People do not use it because it costs them time, not saves them.
Nobody knows when to use it. The tool can do many things. Your team has not been told which of those things matters for their specific role. So they default to not using it, because they are not sure if their use case is the right one.
The feedback loop is invisible. Your team uses the tool. Nothing happens. There is no recognition, no shared learning, no visible improvement in any metric anyone can see. So the use is not reinforced, and it slowly fades.
The exemplars are missing. Your team has not seen someone they respect using the tool well in a real situation. They have read documentation. They have watched demo videos. They have not seen a colleague get a real result on a real task. The most powerful adoption signal is missing.
The tool was chosen for them, not with them. The decision to buy the tool was made by leadership or IT. The team is the receiving end of the rollout, not a participant in the choice. So there is no ownership, no investment, and no real commitment to making it work.
Fix these five and adoption looks completely different.
The 7 tactics that actually drive adoption
This is the practical playbook we use at ThinkSwift when we work with operations teams who have paid for the tool and cannot get the team to use it. These are not theoretical. They are the moves the businesses that have cracked AI adoption are actually making.
Tactic 1. Integrate the tool into the workflow, not next to it
The single biggest determinant of whether your team uses an AI tool is whether using it is faster than not using it.
This means the tool needs to live inside the workflow your team already runs, not as a separate destination they have to navigate to. If your team is already in Slack all day, the AI needs to work in Slack. If they live in your CRM, the AI needs to surface in the CRM. Every extra click, every extra context switch, every separate login is friction that destroys adoption.
The practical exercise is to sit with the team and map exactly where in their existing workflow the AI should appear. Then either configure the tool to work there, or pick a different tool that does. Tools that require the team to go to a separate website, log in separately, and navigate to a feature they need to remember exists will not get used.
This is why so many businesses with ChatGPT enterprise subscriptions report low adoption. The tool is excellent. It just lives outside the workflow. The teams that get adoption with it are the ones that have built it into Slack channels, Chrome extensions, or other places the team is already working.
Tactic 2. Define specific use cases for each role
"Use AI more" is not a directive your team can act on. "Use AI to draft your first version of every client follow-up email" is.
The teams that get high adoption have done the work of defining specific, named use cases for each role in the business. Not generic capabilities. Specific applications.
For an account manager, the named use cases might be: drafting client follow-up emails, summarising call notes into action items, preparing weekly client reports, generating proposal sections. For an operations coordinator, they might be: triaging incoming requests, drafting initial responses to common questions, summarising weekly metrics, capturing meeting notes.
The list does not need to be long. Four to six specific use cases per role is usually right. The point is that every person on the team knows, without thinking, what they should be using AI for in their daily work. The decision of "should I use AI for this?" disappears, because the answer has already been decided for the specific tasks that matter.
This is unglamorous work. It is also the most leveraged hour an operations leader can spend on AI adoption.
Tactic 3. Make the exemplars visible
Your team needs to see someone they respect using AI well on a real task. Not a demo. Not a case study. A real colleague, getting a real result, in a way they can see and understand.
The mechanism for this is straightforward. Identify the early adopters on your team, the people who are already using AI well. Give them visible platforms to share what they are doing. Five-minute team meeting demos. Slack threads with screenshots. Lunch-and-learn sessions where they walk through a real workflow.
The reason this works is that adoption is social. Watching a colleague save thirty minutes on a task you do every day is far more persuasive than any amount of leadership messaging. Seeing the actual output, the actual prompt, the actual saved time, removes the abstraction that keeps most people from trying.
At Roblox, internal product leaders deliberately share their AI workflows in Slack and one-on-ones, sometimes screen-sharing AI output during meetings to solve problems with their colleagues in real time. The visibility is the point. When AI use is hidden, adoption stays low. When it is visible, it spreads.
Tactic 4. Build peer champions, not centralised training
Centralised training programmes get poor results for AI adoption. The reason is that AI use is contextual. The way an account manager uses AI is different from how a finance analyst uses it, and a single training session cannot serve both well.
Peer champions work better. Identify two or three people in each functional area who are early and skilled adopters. Give them explicit roles as the AI champions for their teams. They run small, regular sessions, answer questions, share what they have learned, and create the context-specific guidance the team actually needs.
This is much cheaper than corporate-led training. It is also dramatically more effective, because the champion knows the team's work, the team's context, and the team's specific friction points.
The investment to make this work is small. Some time set aside for the champions to develop their own skill. Some explicit recognition that this is part of their role. Some support from the operations team in terms of materials and shared examples. The return is significant and compounds over time.
Tactic 5. Build feedback into the operating rhythm
Most AI adoption efforts have a launch and then nothing. The team gets the tool. They are told to use it. There is no structured moment after that for surfacing what is working, what is not, and what should change.
The fix is to build AI usage into the operating rhythm of the business. A standing item in team meetings: what AI use saved you time this week. A monthly review where the operations leader looks at usage patterns, picks out wins, and surfaces problems. A quarterly check on which use cases are sticking and which are not.
This is not about surveillance. It is about making the use of AI visible and continuously improvable. When the team sees that their input affects how the tool is rolled out, that wins are shared, and that problems get addressed, the use compounds. When they see nothing, the use fades.
Tactic 6. Connect adoption to performance, but carefully
Some businesses are starting to tie AI adoption to performance reviews. Shopify, famously, asks employees to rate colleagues on a one-to-five scale for how well they "reflexively use AI tools." This is one of the strongest signals possible about what the business expects.
This approach works, but only under specific conditions. The team has to have had time to develop the skill. The use cases have to be clearly defined. The training and tools have to actually be in place. If you tie adoption to performance before any of these conditions are met, you generate compliance and resentment, not real adoption.
A more sustainable version, for most growing businesses, is to make AI fluency a development area in performance reviews. Not a metric you score people on, but a topic of discussion. What AI tools are you using. What use cases have you developed. What would help you do more. This creates the pressure of expectation without the toxicity of mandatory adoption.
Tactic 7. Remove the procurement friction
This is the tactical move most operations leaders overlook. If your business has long approval processes for new AI tools, your team is probably using AI tools you do not know about, from their personal accounts.
Boston Consulting Group's October 2024 survey of one thousand C-suite executives found that seventy-four percent of companies struggle to scale value from AI, and roughly seventy percent of the challenges stem from people and process issues. A meaningful slice of that is procurement friction that pushes legitimate AI use into the shadows.
The fix is to cut the red tape. Create a clear, fast process for the team to request AI tool trials. Give each function a small AI experimentation budget they can use without escalation. Build a simple internal list of approved tools so the team knows what they can use without asking. Duolingo gave every employee three hundred dollars to try AI tools, courses, and subscriptions. Zapier assigned a lead PM specifically to fast-track AI tool approvals through procurement and legal.
The goal is to make legitimate, sanctioned AI use easier than unauthorised AI use. If sanctioned use is hard, your team will continue to use AI in ways you cannot see, support, or measure.
What good adoption looks like in practice
When these tactics are working in a business, the change is visible.
The team is using AI without being asked to. They have specific use cases they reach for instinctively. They are sharing what works in their team channels without prompting. The early adopters are pulling others along. New tools get adopted within weeks rather than quarters because the operating rhythm is already in place. AI use shows up in performance conversations not as compliance but as development.
Most importantly, the time-saved is real and visible. You can see it in cycle times, in capacity freed up, in customer outcomes. The investment in the tools is not theoretical anymore. It is showing up in the business.
This is what AI adoption actually looks like when it works. It is not a one-off project. It is an operating capability that compounds.
The bigger picture
Getting your team to actually use AI tools is not a training problem, a communication problem, or a leadership problem in isolation. It is an operational design problem.
The tools need to live inside the workflow, not next to it. The use cases need to be specific, not general. The exemplars need to be visible. The feedback needs to be structured. The procurement needs to be frictionless. The expectations need to be clear without being punitive.
None of this is complicated. It is also not optional. The businesses that put these tactics in place are seeing genuine AI adoption and the productivity gains that come with it. The ones that are not are paying for subscriptions that nobody uses, running training that nobody applies, and wondering why the AI program is not delivering.
The good news is that the tactics work fast. You do not need a year-long change programme. Most of the moves above can be in place within a quarter. The team will start using the tools, the use will compound, and the business will start to feel the benefit the investment was always supposed to deliver.
The tools are not the problem. The work around the tools is.
