There’s a persistent myth in the AI conversation that goes something like this: to work with AI in any serious way, you need to know how to code. You need a background in statistics. You need to understand machine learning at a level that took a computer science degree to develop.

This narrative is wrong, and it’s actively harmful to the businesses and teams it discourages.

The realistic starting point for most teams is Excel. Maybe some manual reporting. Perhaps a basic Power BI dashboard someone built a year ago that nobody fully understands. And from that starting point, there is a clear, practical path to genuine AI capability. It doesn’t require a CS degree. It doesn’t require hiring a data science team. And it doesn’t require anyone to reinvent themselves over a weekend.

Here’s what it actually looks like.

Where most teams actually are

Before mapping a path forward, it helps to be honest about where most teams begin.

The average knowledge worker spends a meaningful portion of their week in spreadsheets. They’re doing calculations manually that could be automated, building reports that someone else then reformats, and making decisions based on data they’ve pulled together in a process they couldn’t fully document if asked to.

This isn’t a criticism. Spreadsheets are genuinely powerful tools and the fact that most teams can use them is a real foundation to build from. But it does mean that the starting point for most upskilling journeys is much earlier than AI vendors and conference talks suggest.

Acknowledging where teams actually are is the first step to building a programme that works.

The skills ladder, honestly

Think of data and AI capability as a ladder. Each rung builds on the one below it. You don’t have to reach the top rung to create real business value, and not every person on your team needs to climb to the same height.

Rung 1: Structured Excel and data thinking

This is where most teams start. The goal at this stage isn’t new software. It’s better habits with the tools people already use: structured data, consistent formatting, formulas that actually document what they’re doing, and the beginning of analytical thinking around business questions.

Teams that develop this foundation build reports that are accurate and repeatable, not reports that work until someone changes a cell and the whole thing breaks.

Rung 2: Power BI and business intelligence

This is the first major step change. Power BI takes the data that was living in spreadsheets and makes it visible, interactive, and shareable across the organisation. Decisions that used to require a request to the “data person” can now happen because the relevant manager can look at the dashboard themselves.

The business value here is direct: faster decision-making, better visibility, fewer hours spent formatting reports. Most teams can reach a productive level of Power BI capability in a few weeks of structured learning. And Power BI is genuinely just the start — there is a clear progression from dashboards to AI deployment that most teams never complete.

Rung 3: SQL and data querying

SQL is the point at which people start understanding data at the structural level. It’s not about writing clever queries. It’s about being able to answer a business question directly from the underlying data rather than waiting for a report that might or might not exist.

A marketing manager who can write basic SQL queries can answer their own questions about customer behaviour in minutes instead of days. A finance person who understands SQL can validate the numbers in any report rather than taking them on faith.

This is also the rung where people develop a much better intuition for what “data” actually is, how it’s structured, where it lives, and why data quality matters. That intuition becomes extremely important as AI tools enter the picture.

Rung 4: Python for analysis

Python gets over-emphasised as an entry point, but it belongs here, once the foundations are solid.

What Python actually enables at this stage isn’t machine learning or model building. It’s automation of the repetitive analytical tasks that have been eating hours. Pulling data from multiple sources, cleaning it, running standard calculations, and producing a report that used to take a morning can now happen in seconds.

For many people, Python is also the point where working with AI tools starts to feel natural, because they’re already thinking in terms of programmatic logic.

It’s worth being realistic about the time investment here. Learning to be genuinely productive in Python for business analysis takes months, not a weekend. Structured learning with real business use cases as the curriculum accelerates this significantly.

Rung 5: AI concepts and working with agents

This is the rung that most people are trying to jump to directly, often without the foundations below it.

Understanding AI at a practical business level means knowing what different types of AI tools do and don’t do well, how to prompt AI systems to get useful outputs, how to evaluate whether an AI agent is performing correctly, and how to integrate AI capability into real workflows.

None of this requires knowing how to train models or understand neural network architecture. But it does require the analytical thinking and data intuition that the rungs below develop.

Teams at this rung can work with AI tools in a fundamentally different way. They brief AI agents more precisely. They spot errors. They know when to trust the output and when to verify it. They can build workflows that connect AI tools to the data and systems that give them context.

You don’t need to climb every rung

This is an important point that often gets missed.

Not every person on your team needs to reach rung five. And reaching rung two or three already creates significant business value.

A finance team that reaches solid Power BI capability will produce better reporting and make faster decisions than they did before, even if nobody on the team ever writes a Python script. An ops manager who understands SQL will ask better questions of the data, catch more errors, and depend less on the data team for analysis, even if they never work directly with AI agents.

The goal of a team upskilling programme isn’t to turn everyone into a data scientist. It’s to raise the data capability floor across the organisation so that data-informed decisions become the default rather than the exception. The evidence on what that actually unlocks is compelling: data-literate companies significantly outperform their peers on growth and customer acquisition.

The time investment, realistically

Honest estimates matter here, because unrealistic timelines lead to abandoned programmes.

Reaching productive Power BI capability from an Excel baseline: 4 to 8 weeks of structured learning with consistent practice time built in. Not a weekend.

Becoming comfortable with SQL for business analysis: 2 to 4 months of regular practice on real data.

Getting to genuinely useful Python proficiency for business analysis: 4 to 6 months minimum, more realistically 6 to 9 months to feel truly comfortable.

Building AI literacy and the ability to work effectively with AI agents: this one depends heavily on the rungs below it, but with the right foundation, a few months of structured learning gets most people to a productive level.

The common thread is “structured learning.” Self-directed, pick-up-when-I-can approaches to skill development work for some people but fail most. The teams that build real capability do so through structured paths with clear milestones and, ideally, some accountability mechanism.

How EDNA Learn structures this progression

EDNA Learn’s curriculum is designed around this exact ladder. Not as a series of isolated courses people pick from a catalogue, but as structured learning paths that build from wherever a team is starting.

The Power BI pathway is the most-used entry point for business teams. It moves people from their first interaction with the tool through to building complex, production-quality reports and understanding the data model behind them.

From there, structured paths lead into SQL, Python, and AI fluency. Each path is designed with business use cases as the curriculum, not abstract exercises. A professional learning SQL through EDNA is writing queries against data structures that look like the ones in their actual business.

Mentor AI is part of the learning experience, a guide that helps members work through problems, get unstuck, and understand concepts in context. It’s the difference between hitting a wall and stopping, and hitting a wall and finding a way through.

The team effect

One thing that’s consistently true about data upskilling is that the impact is non-linear. A single data-literate person in a department lifts the entire team’s capability.

They build the reports others rely on, but they also model a different way of thinking about problems. They ask questions like “what does the data tell us?” that start to shift how meetings go. They normalise using data to make decisions rather than deferring to whoever has the strongest opinion.

When two or three people in a department have this capability, the culture starts to shift. And when most of the team has it, the kind of AI deployment that actually creates a competitive advantage becomes possible, because you have people who can work intelligently with the tools, not just people who have access to them.

That’s the destination. And for most teams, the path there starts with building better habits in the tools they’re already using every day.

Start your team’s learning path on EDNA Learn

If you’re interested in what the full picture looks like — learning plus deployment — the AI won’t replace your team post explains the philosophy behind why both matter.