How MyDataWork Applies AI to Real Data Work
Most software that advertises AI starts with a chat box and leaves the rest to you. MyDataWork has an AI help assistant too, because users should be able to get guidance wherever they are in the app. But the assistant is not the main AI story. It is the help layer.
The deeper AI story is what happens when AI is built directly into the workflow itself.
That distinction matters. A chat assistant can help you understand the product: where to go, what a feature does, or how to complete a task. MyDataWork’s workflow AI helps you understand the work the product manages: what your assets support, where use cases are incomplete, where leverage exists, what needs attention, and how the estate is performing.
Instead of treating AI as one general-purpose assistant bolted onto the side, MyDataWork applies AI at specific moments in the data practitioner’s workflow. Each function has a clear job: propose a use case, improve one, find leverage, surface issues, or assess the overall estate.
The result is not one AI trying to do everything. It is a help assistant plus five focused workflow functions — each useful on its own, and more powerful together.
Two design choices make that possible.
First, every AI function reads from the same foundation: the structural picture of your work that MyDataWork builds from your assets. That includes things like asset names, file types, systems, schemas, relationships, lineage, linked use cases, and the outcomes you are tracking.
It is context about the work, not the underlying business data. MyDataWork is designed around metadata and structure, not row-level records, cell values, or private business content being sent to an AI service.
That matters because the output is grounded in your actual estate without exposing the sensitive data inside it.
Second, the functions compound. Each one is useful by itself, but they are designed to build on one another. A set of files can become a use case. A use case can become a better-defined initiative. A set of initiatives can reveal reuse, automation, enrichment, or modernization opportunities. The workspace can then surface what needs attention, and the estate assessment can synthesize the whole picture.
Here is how the five functions fit together.
1. Propose: turn assets into use cases
For many data practitioners, the hard part is not doing the work. It is documenting what the work is for.
You may have spreadsheets, Alteryx workflows, dashboards, SQL scripts, notebooks, or other assets that support important decisions. But turning those assets into a clear use case with an objective, outcome, and business purpose is often skipped.
That is where Propose comes in.
You select a set of assets, and MyDataWork proposes use cases based on what those assets appear to support. Instead of starting with a blank form, you start with a grounded draft: a suggested use case name, objective, and purpose based on the structure and relationships of the assets you selected.
You stay in control. You can edit, accept, reject, or refine what MyDataWork suggests. The goal is not to replace your judgment. The goal is to get the first version of the work into a structured form quickly.
Why it matters: Propose helps turn raw assets into documented business work. That is the starting point for everything else.
2. Recommendations: make a use case stronger
Once a use case exists, it can still be thin.
It may have a name but not enough detail. It may have linked assets but unclear gaps. It may describe the work but not explain what needs to improve.
Recommendations helps strengthen that use case.
It looks at the use case description and the assets linked to it, then suggests ways to make the use case more complete. That might mean identifying missing context, suggesting connections worth making explicit, pointing out gaps, or recommending improvements that make the use case easier to act on.
This works whether the use case was generated by Propose or created manually. The important point is that Recommendations reads from the same grounded context: the use case, its assets, and the structure around them.
Why it matters: Recommendations turns a recorded use case into a more complete and actionable one.
3. Leverage: find where the opportunity is
Once use cases are documented, the next question is simple:
Where is the leverage?
The Leverage tab answers that question from four angles. Each one looks at your use cases and assets through a different strategic lens.
Find reuse opportunities looks inward. It asks where work overlaps across use cases, projects, clients, or teams. If the same assets, topics, or patterns are appearing in multiple places, there may be an opportunity to reuse, consolidate, or standardize the work instead of rebuilding it.
Identify automation candidates looks at process. It asks which parts of the work appear manual, repetitive, or ready for automation. The result is not just a vague suggestion to automate. It includes the assets involved, the use case affected, estimated value, effort, and a concrete next step.
Discover marketplace data looks outward. It asks whether external datasets could enrich the use case. For example, a demand forecasting or inventory planning use case might benefit from weather data, retail sales data, logistics data, consumer behavior data, or supply chain data available through cloud marketplaces.
Migration Assist looks at structure. It asks whether part of the work could be modernized: an Excel model moved to Python, an Alteryx workflow rebuilt in dbt, or another tool transition that could improve maintainability, scale, governance, or cost.
Together, these four lenses form a practical opportunity map:
- Reuse what you already have.
- Automate what is still manual.
- Enrich the work with data you do not have.
- Modernize the tooling where it makes sense.
Why it matters: Leverage helps move from documentation to strategy. It shows where the current estate may contain the next improvement, project, or investment case.
4. Workspace Agent: surface what needs attention
The first three functions are things you run when you are deliberately working on assets or use cases.
The Workspace Agent is different.
It acts like a standing worklist for your workspace. Run it when you want to know what needs attention. It scans across your assets and use cases and surfaces specific findings that are worth acting on.
The Workspace Agent checks six conditions across three categories.
Cleanup looks for issues like use cases missing stakeholders or work that has gone stale.
Activity looks for changes, such as newly added assets that have not yet been linked to a use case, or removed assets that are still referenced elsewhere.
Insight looks for higher-value signals, such as use cases tracking below estimate or assets that quietly support many different use cases — the hidden infrastructure of your workspace.
The detection itself is rule-based and deterministic. AI is used to explain the finding in clear language, not to invent the finding. Each item points back to the underlying asset, use case, or condition that caused it to appear.
That distinction matters. The Workspace Agent is not a free-form chatbot making guesses. It is a detector that turns real workspace conditions into a readable worklist.
Why it matters: The Workspace Agent gives you a practical answer to the question, “What should I look at next?”
5. Asset Estate Assessment: synthesize the whole picture
The Workspace Agent gives you discrete findings.
The Asset Estate Assessment gives you the broader read.
It is a point-in-time assessment of the health and readiness of your asset estate. It consumes the Workspace Agent’s findings, adds analysis of your lineage and use cases, and produces a narrative report that explains what is documented, what is missing, what is at risk, and what may be ready for the next step.
The assessment is organized around the questions a senior analyst, analytics manager, consultant, or data leader would naturally ask:
What assets are in the estate?
Which use cases do they support?
Where are the important connections?
Which dependencies are not yet documented?
Where are there opportunities for reuse, automation, consolidation, or better ownership?
What should be done next?
This is different from the Workspace Agent.
The Workspace Agent is a detector. It gives you specific items to act on.
The Asset Estate Assessment is a synthesis. It gives you a broader judgment about the estate as a whole.
Both are useful, but they serve different purposes. One helps you manage the running worklist. The other helps you understand the overall condition of the workspace and communicate it to others.
Why it matters: The Asset Estate Assessment turns the accumulated context in MyDataWork into an executive-ready readout of where things stand and what should happen next.
Why the five functions work better together
Each AI function in MyDataWork is useful on its own.
Propose helps you create use cases from assets.
Recommendations helps improve those use cases.
Leverage helps identify strategic opportunities.
The Workspace Agent helps surface what needs attention.
The Asset Estate Assessment helps synthesize the whole estate.
But the real value comes from the way they build on each other.
A spreadsheet, workflow, or dashboard is not very useful as an isolated object. Its value comes from what it supports: a forecast, a planning process, a customer analysis, a supply chain decision, a financial model, a migration effort, or an operational workflow.
MyDataWork’s AI functions are designed around that reality. They do not start with a blank chat prompt. They start with the structure of the work itself.
That is why the outputs become more useful as more of your work lives in MyDataWork. The system can see more assets, more relationships, more lineage, more use cases, and more outcomes. Each function has more context to work from.
The same design also makes scope important.
You do not have to run these functions against your whole workspace. You can scope them to a client, project, team, domain, or group of use cases. That is especially useful for consultants, agencies, analytics teams, and data leaders who need to keep different bodies of work separate.
- A consultant can run Propose and Leverage against one client’s assets.
- A data leader can run an Estate Assessment for one domain.
- An analyst can use Recommendations to strengthen one use case before sharing it with a manager.
- A team can run the Workspace Agent periodically to keep their work organized and current.
That is the throughline.
MyDataWork does not use AI as a single chat box that tries to answer anything. It uses AI in five focused ways, each tied to a real step in managing data work.
It helps you define the work.
Improve the work.
Find opportunity in the work.
Monitor the work.
Assess the work.
And it does all of that from the same grounded context: your assets, your lineage, your use cases, and your outcomes — not generic advice, and not your underlying data values.
That is what makes the AI useful.
It understands the shape of the work well enough to help you act on it.