My Take on Redesigning Work Through AI and Automation

AI is reshaping how companies operate — but not in the way most people imagine. While the technology is advancing rapidly, the real story is how humans, teams and organisations adapt around it. Over the past year, through my work with startups, scaleups and enterprise teams, I’ve seen a pattern emerge: the companies that benefit most from AI are not the ones with the biggest budgets, but the ones who learn how to redesign their workflows.

A futuristic AI interface showcasing product analytics and insights.

What surprised me recently is that many large enterprises are starting to give employees access to all major AI tools at the highest subscription tier and simply telling them: “Figure out how to make yourself more effective.” It’s bold. It’s messy. And it tells us something important:AI value does not come from the tool — it comes from how people learn to use it.

Startups and scaleups, in many ways, are better positioned to take advantage of this. They have fewer layers, faster decision cycles and more freedom to redesign work from scratch. But they still need a structured way to think about AI and automation. Below is a practical framework based on what I’m seeing across real operations.

Why Enterprises Struggle and What Startups Can Learn From It

Across dozens of conversations with leaders in manufacturing, tech, e-commerce, academia, and consulting, five themes consistently surface:

• Leadership drives adoption, not tooling. AI-positive individuals create the momentum.

• Inertia slows everything down — processes built years ago resist change.

• Trust is uneven, both in data protection and in the consistency of AI outputs.

• Startups start with AI; enterprises try to retrofit it — and the friction shows.

• Efficiency gains are hard to quantify. AI shifts work, but doesn’t erase it.

Enterprises are experimenting — sometimes very effectively, sometimes in fragmented ways — but true transformation will take time. This is where smaller companies have an advantage: you can redesign workflows today, not next year.

A Practical Framework for Founders, Operators and Teams

Instead of asking “What can AI do?”, start with:What does my work actually consist of? And which parts of it can I automate or augment?

Break your activities down and evaluate each one through these questions:

1 — Fully Online, Repetitive, Clearly Defined Tasks → Use AI Generated Code

If a task is:

• digital,

• highly repetitive, and

• follows a predictable structure,

then it’s a strong candidate for AI-written scripts.Ask an AI model to generate the code, test it, refine it and then use it whenever needed.

This is ideal for: parsing files, tagging content, cleaning data, renaming assets, transforming formats — all the “micro-operations” teams waste hours on.

2 — Tasks Triggered by Events, With Data Moving Between Tools → Use an Automation Platform

If your work starts with a trigger (“new lead”, “new request”, “new order”) and always flows through the same steps, then an automation tool (n8n, Zapier, Make) is the right choice.

Startups in particular should systemise these flows early — every manual step compounds over time.

3 — Tasks With a Trigger Plus a Need for Reasoning → Add an AI Agent Into the Automation

Some work requires:

• research

• summarisation

• categorisation

• editorial judgment

These tasks are perfect for AI agents embedded in your automation.You define the steps, the agent handles the interpretation.

This bridges the gap between rigid automations and messy human tasks.

4 — Tasks That Should Not Run Unchecked → Add Human in the Loop Controls

If an automated process produces an output that must be reviewed — for accuracy, tone, financial risk, or brand safety — add a checkpoint.

Let automation prepare everything, then send the output to you via Slack, email, or a dashboard for approval.Once you confirm, it continues.

This preserves speed while maintaining judgment.

5 — Work Requiring Many Searches or Comparisons → Use Search AI Tools Like Perplexity

If a task requires collecting information from multiple sources, comparing insights, or synthesising findings, a search-specialised AI tool is the fastest route.

If repeated often, wrap it in an automation with scheduled triggers.

6 — One-Off, High-Level Work → Use a strong AI model like ChatGPT as your analyst

For deep research, option evaluation, recommendation-making or strategic planning, use an AI model as your assistant.

If the domain is specialised:

• create a project,

• upload key documents,

• set system instructions (tone, role, constraints),

• and anchor the model in your context.

This turns AI from a generic assistant into a tailored expert.

Where This All Leads

Startups and scaleups that adopt AI early will operate with:

• leaner teams,

• faster delivery cycles,

• better customer responsiveness,

• higher-quality decisions, and

• better internal knowledge systems.

Enterprises will get there too — but slower, because culture, structure and legacy systems take time to shift.For now, AI transformation is still shaped more by people than platforms.

The Opportunity

If you’re building or scaling a company today, AI and automation aren’t simply tools — they’re leverage.They compress the cost of experimentation, reduce operational drag, and let you punch above your weight.

The future will belong to teams who:

• understand their workflows,

• separate the repetitive from the analytical,

• automate where possible,

• and use AI agents to amplify the rest.