Most operations managers run change initiatives with the wrong set of tools for each phase.
They use the same general project management software to plan a CRM rollout as they do to track office supply orders. They write stakeholder emails manually while the same AI that drafts their meeting summaries sits idle. They declare the change "done" when the software goes live, then wonder why three months later half the team is still using the old process.
The problem isn't a lack of technology. It's a lack of clarity about which AI tool does what job at which point in the change process.
This guide maps specific AI tools to each phase of the ADKAR model — the most widely used change management framework from Prosci. If you've never heard of ADKAR, that's fine: by the end you'll understand it through the lens of tools you can start using this week.
Why ADKAR — and Why It Matters for Tool Selection
ADKAR stands for: Awareness → Desire → Knowledge → Ability → Reinforcement.
Each letter represents a milestone an employee must reach before a change actually sticks. The sequence matters: you can't train people on a new process (Knowledge) if they don't yet want to change (Desire). And you can't sustain a change (Reinforcement) if people never fully developed the skills for it (Ability).
Most change initiatives fail not because the change was wrong, but because they skipped phases. They announced the change (Awareness), ran one training (Knowledge), and skipped Desire and Ability entirely.
AI tools help you stop skipping phases by making each phase faster and cheaper to execute. Here's how.
Phase 1 — Awareness: AI for Impact Assessment and Communication Planning
The job: Tell the right people, in the right way, that something is changing — and why.
This is where most ops managers spend the least time and should spend more. A poorly communicated change that a single stakeholder misunderstood can derail months of work.
Tools for this phase
ChatGPT or Claude (free / $20–$25/user/month)
Use generative AI to draft your initial impact assessment and stakeholder communication plan. A prompt like:
"We're migrating from Salesforce to HubSpot. The change affects 40 salespeople and 5 sales managers. Timeline is 8 weeks. Main concerns will be around data migration and relearning workflows. Write a stakeholder impact assessment and a 4-week communication plan with email templates for Week 1 and Week 4."
Takes 5 minutes. Produces a usable skeleton. You edit for tone and org-specific context — you don't write from scratch.
Claude's free tier at claude.ai handles this well. ChatGPT's free tier works too. For teams, ChatGPT Team ($25/user/month) and Claude Team ($25/user/month) add shared workspaces, but for Awareness-phase work, the free versions are sufficient.
Notion AI (included in Business plan — $20/user/month billed annually)
Create a living change management knowledge base in Notion. The AI helps you generate and iterate on:
- Stakeholder maps
- FAQ documents for the change initiative
- Meeting notes from steering committee sessions
- A single "source of truth" page employees can reference
The advantage of Notion over a Google Doc is that AI-assisted editing, search, and summarization are built in. When your change initiative spans 12 weeks and generates 40 documents, Notion AI finds what you need in seconds.
For a broader look at AI-assisted employee communication — beyond change initiatives — see our guide to AI internal communications tools.
Phase 2 — Desire: AI for Sentiment Analysis and Resistance Identification
The job: Understand who wants this change, who doesn't, and why.
This is the phase that feels most "soft" and is therefore most often skipped. It's also the phase where AI gives you the clearest ROI.
Tools for this phase
Leapsome (from ~$8/user/month)
Leapsome runs structured employee surveys with AI-powered analytics. For a change initiative, you can:
- Deploy a 3-question pulse survey mid-week asking about readiness and concerns
- Let the AI surface common sentiment themes from open-text responses
- Identify teams or managers where resistance is concentrated
You're not guessing at resistance. You have data.
Leapsome requires a minimum $6,000 annual contract, which makes it a mid-market fit. For smaller teams (under 30 people), you can approximate this with a well-designed Typeform survey and ChatGPT analysis of the text responses — not as elegant, but functional.
ChatGPT or Claude (same cost as above)
Feed the open-text responses from any survey tool into ChatGPT or Claude and ask: "Categorize these 40 employee responses into themes. Which concerns appear most frequently? Which are one-offs? What questions haven't been answered yet?"
This costs $0 in additional tooling if you're already using these tools for Phase 1.
Phase 3 — Knowledge: AI for Training Content and SOPs
The job: Ensure people know exactly what they need to do differently after the change.
This is where most ops managers invest the most effort — and where the most time gets wasted creating content that could be AI-assisted.
Tools for this phase
Loom with AI features (free plan available; Business plan ~$12.50/user/month)
Loom lets you record async screen-capture training videos. The AI features do the heavy lifting post-recording:
- Auto-generates a title and summary
- Creates chapter markers for long recordings
- Produces a searchable transcript
- Generates follow-up tasks from the video content
For a CRM migration, instead of scheduling 8 live training sessions, you record once. The AI-generated summary and chapters mean employees can watch only the sections relevant to their role.
The free plan limits video length to 5 minutes (sufficient for most how-to clips). The Business plan removes length limits and adds advanced AI features.
Claude or ChatGPT for SOP generation
A prompt like: "Write a step-by-step SOP for a salesperson logging a deal in HubSpot. The audience has never used HubSpot before. They're used to logging deals in Salesforce with these fields: [list fields]. Format it as a numbered list with screenshots cues noted in [brackets]."
Produces a first draft in 90 seconds. You add context and screenshots. You do not start from a blank page.
For more structured SOP workflows, see our guide to AI SOP generators — these tools go deeper on documentation automation.
Phase 4 — Ability: AI-Assisted Digital Adoption Platforms
The job: Help employees do the new thing in the moment they need to do it, not in a training session two weeks earlier.
Knowledge and Ability are different. You can know how to log a deal in HubSpot and still freeze when you see an unfamiliar screen on day one. Ability is built through doing — and digital adoption platforms (DAPs) are built exactly for this.
Tools for this phase
Whatfix (custom pricing — quote required; typically $10,000+/year for mid-market)
Whatfix overlays your existing software with in-app guidance: step-by-step walkthroughs, tooltips, smart alerts when an employee is stuck on a step. The AI component:
- Identifies where users are dropping off or asking for help most
- Surfaces guidance proactively at the right moment
- Generates adoption analytics by team, role, and feature
Whatfix is priced for teams of 100+ and requires a sales call for pricing. It's not a tool you spin up in a week — implementation takes 4–8 weeks with a proper rollout. But if you're managing a change that affects software used by 100+ people for years to come, the ROI is real.
WalkMe (custom pricing — enterprise-focused)
WalkMe is the larger, more established competitor to Whatfix. Also enterprise-priced, also requires a demo. The key differentiator: WalkMe has a longer track record in complex enterprise software environments (SAP, Salesforce, Workday). If you're running a Salesforce-to-HubSpot migration for a 500-person team, WalkMe has more case studies in that environment. For SMBs or mid-market, Whatfix is typically easier to implement.
Note on cost: If your budget doesn't extend to Whatfix or WalkMe, Loom recordings embedded directly in the application (as linked resources or shared via Slack at the moment of use) approximate the "just-in-time help" behavior without the platform investment.
Phase 5 — Reinforcement: AI for Adoption Tracking
The job: Confirm the change is actually happening. Catch backsliding before it becomes the norm.
This phase gets skipped more than any other. You declare the project "done" when go-live happens, not when adoption is sustained. Six months later, 40% of your team has reverted to the old process and nobody has the data to prove it.
Tools for this phase
Leapsome (same as Phase 2)
Run monthly pulse surveys post-go-live. Track the trend. Leapsome's AI surfaces whether adoption sentiment is improving or declining — and flags teams where it's declining before it becomes invisible.
ChatGPT or Claude for analytics interpretation
If you're pulling usage data from your new software (HubSpot, Notion, whatever changed), paste it into Claude with the question: "Here is weekly active user data for our HubSpot rollout for the past 8 weeks. Identify which teams show the strongest adoption trend and which are declining. Suggest 2–3 targeted interventions."
You don't need a BI tool. You need data and a prompt.
For a comprehensive view of AI tools across the entire operations function — beyond change management — see our roundup of best AI tools for operations.
Comparison Table
| Tool | ADKAR Phase | Price | Best for | Limitation |
|---|---|---|---|---|
| Claude (free tier) | A, D, K, R | Free | Drafting comms, SOPs, survey analysis | No persistent memory; you re-explain context each session |
| ChatGPT (free tier) | A, D, K, R | Free | Same as Claude; slightly better for structured templates | Same context limitation |
| Notion AI | A, K | $20/user/month (Business) | Living change knowledge base, collaborative docs | AI quality is good but not as strong as dedicated LLMs for generation |
| Loom (AI features) | K | Free / $12.50/user/month | Async training videos with AI summaries and chapters | Free plan capped at 5-min videos |
| Whatfix | A (ability) | Custom (~$10k+/year) | In-app guidance and adoption analytics for software changes | Enterprise pricing; 4–8 week implementation |
| WalkMe | A (ability) | Custom (enterprise) | Large-scale enterprise software adoption | Very enterprise-only; overkill for teams under 200 |
| Leapsome | D, R | ~$8/user/month (min $6k/year) | Employee sentiment tracking, pulse surveys | Minimum contract makes it mid-market focused |
How to Get Started Without a Dedicated Change Manager
If you're running this yourself — no HR team, no change management consultant — here's the three-step minimum viable process:
Step 1 — Build your impact assessment in 30 minutes (Week 1)
Open Claude or ChatGPT. Use this prompt:
"I'm leading a [describe change: process change / software migration / team restructure]. It affects [number] people in [departments]. Timeline is [X weeks]. Main concerns I anticipate: [list]. Create a stakeholder impact assessment, a 4-week communication timeline, and a FAQ for employees."
Save the output in Notion (free plan works). Refine with your specific context. This is your change management foundation document.
Step 2 — Create training content in 1 day (Week 3–4)
Record 3–5 Loom videos showing the new process. Let Loom AI generate the transcripts and summaries. Paste the transcript of your most complex video into Claude and ask it to produce a step-by-step SOP. Share both video + SOP.
Step 3 — Track adoption with a weekly 3-question pulse (Weeks 5–10)
Use Typeform (free) to send:
- "On a scale of 1–5, how comfortable are you with [new process]?"
- "What's the biggest obstacle you're facing?"
- "What one thing would make this easier?"
Paste the text responses into Claude weekly. Ask for theme analysis. Act on the top 2 themes.
This workflow costs you $0 in new tools if your team already has Loom (free), Notion (free), and Claude or ChatGPT (free). It takes roughly 4 hours to set up and 30 minutes per week to maintain.
What Most Ops Managers Get Wrong
Three patterns show up repeatedly in failed change initiatives:
1. Using AI to skip the Desire phase.
Generative AI makes it easy to produce beautiful communication plans and training materials. It does not make employees want to change. If you've done Phase 1 (Awareness) and Phase 3 (Knowledge) but skipped Phase 2 (Desire), you have well-informed people who still don't want to do the new thing. No tool fixes that. A conversation does.
2. Treating go-live as done.
The project management mindset — tasks complete, milestone hit, project closed — is the wrong frame for change management. ADKAR's Reinforcement phase starts at go-live, not ends there. Build a 90-day post-launch tracking plan before you launch, not after.
3. Deploying a DAP for a change that doesn't need one.
Whatfix and WalkMe solve a specific problem: complex software that people use frequently and where getting the steps wrong has consequences. If you're managing a policy change, a team restructure, or a new meeting cadence, you don't need an in-app guidance platform. You need better communication and manager coaching. Matching the tool to the problem type is more important than having the best tool.
The One-Week Start
If you manage change initiatives for an operations team and you're not using AI yet, here's the one-week experiment:
Take a change initiative you're currently planning or just kicked off. Spend 30 minutes with Claude or ChatGPT using the impact assessment prompt above. See if the output is useful. If it is, add Loom for training content in Week 3. If it isn't, you've spent 30 minutes and learned that generative AI isn't the right fit for your specific situation.
You don't need to buy anything to start. You need to run the experiment.
Originally published on Superdots.
Top comments (0)