AI Transformation is about creating a Human-Agent Partnership to unlock new levels of productivity and innovation.

Planning your AI Digital Transformation

The Watertm Workforce Transformation Platform offers a clear Roadmap to Automation at scale

The 15 Steps of AI Transformation

1. Inventory Functional Areas: Identify your organization’s functional departments (e.g., Recruiting, Sales, Accounting), or at least those you wish to transform by building a Human-Agent Partnership.
2. Collaborate with Department Leads: Work with supervisors to create a list of repetitive projects performed by their department.
3. Highlight High-Impact Opportunities: Identify which projects, if automated, would deliver significant and measurable value.
4. Task List: Choose the top candidate project and enumerate all tasks required for its completion.
5. Task Decomposition: Break down high-value tasks into microtasks or “Action Items” that would normally be delegated to a single person. This list forms your use case candidate pool.
6. Identify Automatable Action Items: List the Action Items that can currently be automated with existing tech—this forms your A-List.
7. Prioritize A-List Action Items: Use FVC scoring (frequency × value ÷ cost). Consider:
   • • Development/acquisition cost
     • Anticipated repetition frequency of use
     • Anticipated value per use
     • ROI (model it out based on repetition, value and cost)
8. Plan your develop/acquisition strategy: Will you acquire? If develop, which tools will you use? For convenience we may refer to all automations as “Agents, but in truth they may form a mixed bag;
   • • Repetitive Process Automations (RPAs) using tools like n8n, Make.com
     • MCP servers/tools (narrow utilities/services)
     • True Reasoning Agents capable of qualitative decisions
9. Develop/acquire Automations.
10. Integrate into test workflows: Selectively grant team leads or test users access to use the Agents, ensuring security and appropriate controls. Refine instructions and prompts until they produce consistent results; store effective prompts in a unified library, for future team access.
11. Authorize Wisely: Selectively grant production access to authorized employees.
12. Update SOPs: Update SOPs, noting which Action Items will be performed by Agents.
13. Operationalize Workflows: Train and motivate employee adoption of new SOPs.
14. Measure Impact: Track labor hours saved, cycle time reduced, and AI costs; calculate ROI. Use adoption-gap reports to spot underutilized AI areas and adjust priorities accordingly.
15. Expand Automations: Return to step 9 to develop the next set of automations as the curret batch of A-List items are put into play.

Download Transformation Infographic PDF:

Identifying AI Use Cases  with the SOP Method

Many organizations struggle to identify viable automation opportunities that come with the magic combination of rapid deployment, low risk, low implementation costs, and high ROI.

Below we reveal methods not just for identifying use cases, but also for prioritization based on the magic criteria mentioned above.

 The secret to finding hundreds of use cases is to break projects down into tasks, then break those tasks down into the most granular of microtasks. This is how engineerable use cases are magically revealed. If you focus on the grains of sand, you will never run out of use cases. Once automated, the microtasks can be stitched together to form complex workflows, which are the toolkits of agents.

Inside Water, the smallest component of a workflow is referred to as an Action Item, which consists of a single microtask, to be assigned to one person (or Sidekick). It is these narrow Action Items that become your use case candidates, for further evaluation and priortization. This process can be as simple as copying  and pasting existing SOPs into the Water platform. If standard operating procedures do not exist, they must now be created.

Why SOPs are Necessary to an Organization’s Transformation Journey

Your SOPs are the instructions that both humans Sidekick agents can follow to perform the tasks. Without an SOP, a new employee would be dependent on expensive, time-consuming, error-prone, mentor-based training. An SOP is far more repeatable and scalable. The same applies to Sidekicks. There is no need for them to learn through trial-and-error. A simple step-by-step explanation of the workflow (an SOP!) is all they generally need, using the exact same language you would use to train or retrain a human employee.

A common-sense principle of automation is to keep the use case narrow. Narrow is easy. Narrow is reliable. Broad is not! That’s why we focus on automating Action Items rather than complex projects or tasks. Small Action Items are easy and fast to develop and deploy in production. And they are reliable!

The Water platform provides an interface where existing workflows and tasks can be broken down and documented into narrow bits and pieces (“Action Items”). We refer to this planning process as Task Decomposition. If an organization has existing SOP documentation, this decomposition process can be performed by a consultant or a specialist on the team.

Documenting the various components of a larger workflow in the smallest possible components makes it surprisingly easy to identify use cases for AI.

Example of Decomposing a Logo Development Task

Action Item #1 – Plan and document scope of task

Action Item #2 – Gather brand guidelines

Action Item #3 – Generate draft logos

Action Item #4 – Send logos to stakeholders for review

Action Item #5 – Make revisions

Action Item #6 – Get final approvals

Action Item #7 – Organize final files and wrap up task

Next, the transformation leader must review each Action Item, asking themselves “Can this be automated today, with the existing state of technologies, or must this wait for future developments?”

At the time of writing this article in September 2025, we see that Action Items #2, 3, 4  & 7 are all candidates for automation.

Those that can be automated (the “A-List”) today go on a list for evaluation, which will be referred to during assessment and evaluation, using some method of prioritization, such as those methods proposed above, (with some additional methods described below).

Eventually, all Action Items in the world might be automated. This might be ten years from now or perhaps fifty. The timing is uncertain, but we are most certainly on that trajectory. So even if only one in ten microtasks can be automated today, eventually that will become two in ten, then three in ten, and so on.

There is no need to struggle with solving difficult automation challenges today, because there will be plenty of easy ones that reveal themselves using the process described here.

Prioritization Methods

Below you’ll learn about planning methods made easier by using reliable models for prioritization. One method is focused on pre-development planning, the other on post-development planning (i.e. optimization).

The FVC Prioritization Method

This is a pre-development prioritization process, focused on the value of automating vs the cost of automating.

The basic formula is frequency*value/cost. Here is an example:

In this example, automating the task ‘Create daily post’ is clearly identified as the highest priority. 

Note: To make the score more readable, you can modify the basic formula so the score is a product of 1000 or even higher, like this: (f*v)/c*1000

Standardize on units of frequency and value

In the above example, we standardized on a monthly frequency, and a value measured in hours. You may alternatively standardize on a different frequency unit (weekly, daily, etc.), and the unit of value in dollars or per incident, or any other method of valuing the task. The formula f*v/c does not change.

Probability of Failure?

If you believe there is a risk of your future automation process failing due to unforeseen obstacles, such as reliability issues, then the formula can be adjusted to (f*v)/c*p where p=probability of failure.

Including a probability of failure can paint a different prioritization picture.

You may alternatively add a ‘time’ factor, where
t=time to deploy the new automation

However; for simplicity, cost can be a proxy for time, which may be sufficient for most cases.

Getting Task Frequency Data

The FVC method of prioritization requires the identification of tasks or microtasks performed repeatedly in an organization. 

Water uses historic data to provide frequency of every task and microtask across an organization. In the simplest of terms, if multiple tasks have the same name, they are grouped. However; many tasks, while similar in nature, have names which humans have chosen to customize, not always for good reason, but it happens. Water’s algorithm performs vector analysis so grouping take place, even if there are dissimilar task names.

The algorithm goes even further by identifying microtasks using the Action Item structure, which is built into water.

This data is available after an organization has been using Water for a minimum of two weeks. Water uses a familiar architecture to structure workflows in a format that is intuitive for humans, and highly readable by AI.

• Organizations – Contains a collection of related projects (such as those belonging to a client, customer, division, region, ministry or department)
• Projects – Contains a collection of related tasks
• Tasks – Contains a collection of related Action Items
• Action Items – The smallest granular event performed by a single user

Water allows dozens or even hundreds of automation opportunities to be swiftly identified, then narrowed down to those repetitive micro-tasks offering the quickest and largest payoffs.

The Adoption Gap Method

As the organization goes through the transformation process, you’ll get insights into adoption and its impact on productivity at every organizational layer, and for every user. 

Adoption-gap reports reveal which user, team, department or division is utilizing AI to complete work. The ‘gap’ reveals who is lagging behind.

The gap may be a result of suitable AI tools not being available yet. Or it can be a result of low adoption. In either case, the data can be used to reset priorities on future development and training.

Water provides transformation leaders insights into metrics that make planning the AI digital transformation process much faster, with more predictable outcomes.

Technology Decisions

Sidekicks can control agents and workflows developed on any platform, including A2A, MCP or even workflow tools such as N8N or Zapier.

These agents and workflows will be interacting with your data so consider how you will lock it down. Within Water you can designate which of your people will get access, so on this side it is secure.

Securely Roll Out your Agents to the Team

Inspira’s Water Platform has enterprise-grade security, allowing you make your agents available to select team members, while restricting others.

Agent’s are added through a conventent form, including authentication requirements. These newly registered agents become ‘skills’ that can be selectively assigned to one or more Sidekicks.

Prompts that will reliably invoke these new skills can be optimized and stored in Water’s built in prompt libary.