AI TOOLS FOR BUSINESS

Your team's practical guide to the platforms, skills, and workflows that drive real results

The AI tools landscape moves fast. The principles for using them well stay constant. Here's how to cut through the noise and focus on what actually delivers value.

Last updated: March 2026

Who this guide is for

*Last updated: February 2026*

If you're a knowledge worker trying to make sense of which AI tools to invest your time in, or a manager trying to decide what your team should be using, this guide is for you. It's also useful if you're a senior leader trying to understand the practical reality of what these tools actually do, rather than what the vendor pitches promise.

We've spent the past three years working hands-on with every major AI platform, testing them in real business workflows, and helping organisations figure out what actually works. And the single biggest thing we've learned is this: the tool matters far less than how you use it. A mediocre prompt on the best platform will produce worse results than a thoughtful prompt on a decent one. That's good news, because it means the biggest lever for improvement is skill, not spend.

So this guide covers both. The tools worth knowing about and the principles that make them useful. Whether you're just getting started or looking to level up how your team works with AI, there's something here for you.

Contents: what you'll find in this guide

• Chapter 1: The Big Four: ChatGPT, Copilot, Gemini, and Claude
• Chapter 2: AI agents: what they actually are
• Chapter 3: CustomGPTs and building your own tools
• Chapter 4: Choosing the right tool for your context
• Chapter 5: The skills that matter more than the tools
• Chapter 6: Where tools go wrong (case studies)
• Further reading

THE BIG FOUR: CHATGPT, COPILOT, GEMINI, AND CLAUDE

The AI market has consolidated around four major platforms. Each has strengths and each has limitations. And all four are evolving so fast that any specific capability comparison will be outdated within months. So we'll focus on the strategic positioning that's more stable, because that's what helps you make a decision you won't regret.

ChatGPT (OpenAI) remains the most widely known and used. It has the largest user base, the broadest plugin and integration ecosystem, and the most public mindshare. For most people starting with AI, ChatGPT is where they begin. Its strengths are breadth, integration with DALL-E for image generation, and a mature interface that works well for general-purpose tasks. The community around it is also massive, which means there's no shortage of tutorials, templates, and use-case examples to learn from.

Microsoft Copilot has become a world-class option. That's a sentence we couldn't have written a year ago. The early versions were underwhelming. But Microsoft has invested heavily, and Copilot's integration across the Microsoft 365 suite (Word, Excel, PowerPoint, Outlook, Teams) makes it the natural choice for organisations running on Microsoft infrastructure. AI just got Excel and data-work superpowers is a good example of what this looks like in practice. The strategic significance of Copilot is less about the AI itself and more about where it sits. The AI is right there in the document, the inbox, the spreadsheet. No context-switching required. For teams that live in Microsoft 365 all day, that integration alone can be transformative.

Gemini (Google) is where we've been going all in and for good reason. Google's integration of AI across its entire ecosystem (Gmail, Docs, Sheets, Drive, Search) means that for organisations already in the Google Workspace, Gemini becomes part of how you already work rather than a separate tool. The deep integration with Google's infrastructure makes it particularly powerful for data-heavy and research-heavy workflows. And Google's advantage in search and information retrieval gives Gemini a natural edge for tasks that require pulling together information from multiple sources.

Claude (Anthropic) has carved a reputation for nuanced, thoughtful responses, particularly on complex writing, analysis, and reasoning tasks. It tends to be the favourite of people who do a lot of deep knowledge work. Its approach to safety and responsible AI also resonates with organisations that care about governance. Claude's strength in handling long, complex documents and maintaining context across extended conversations makes it especially useful for strategic and analytical work.

You'll notice a pattern with Copilot and Gemini: their biggest advantage is ecosystem integration. If your organisation lives in Microsoft 365, Copilot is the obvious starting point. If you run on Google Workspace, Gemini makes more sense. ChatGPT and Claude are platform-agnostic and strong as standalone tools. The earlier Copilot Pro review captures how fast things move in this space. Any point-in-time comparison has a shelf life measured in months.

The honest take? For most business use cases, all four are capable enough to deliver strong results. The choice should come down to your existing technology stack, your team's comfort level, and what specific workflows you're trying to improve. Don't get paralysed by platform choice. Pick one, get proficient, and expand from there. The organisations getting the best results aren't the ones who picked the "best" tool. They're the ones who picked a good tool and invested in learning to use it well.

AI AGENTS: WHAT THEY ACTUALLY ARE

The term "AI agent" has been thoroughly overloaded by marketing. Every chatbot, every automation tool, every AI-powered feature is calling itself an agent. So let's be direct about what AI agents actually are.

A genuine AI agent is an AI system that can take a goal, break it into steps, execute those steps across multiple tools and data sources, adapt when something doesn't work, and deliver a result. It's not a chatbot with a fancy name. It's a system that operates with some degree of autonomy across a multi-step process. And the capability is moving fast.

Think about the difference between asking an AI to "write me an email" (that's a chatbot) and asking an AI to "research our competitor's latest product launch, summarise the key features, compare them to our product, draft a briefing for the sales team, and schedule a meeting to discuss it." That second one requires planning, tool use, data access, and judgment. That's agent-level work. And it's the kind of workflow that's becoming possible now, not in some distant future.

AI agents are coming was something we wrote in early 2024 when the concept was still emerging. Since then, the reality has arrived faster than expected. OpenAI, Google, Anthropic, and Microsoft are all building agent frameworks. And the practical implications for business are significant. We're moving from AI that answers questions to AI that completes workflows.

What will set you apart when AI agents are everywhere? is the question that matters for anyone thinking about their career or their business strategy. When the agents are doing the execution, the value shifts to the humans who can set the right goals, ask the right questions, evaluate the outputs, and provide the judgment that agents lack. So rather than seeing agents as a threat, the opportunity is in becoming the person who knows how to direct them well.

74% of companies plan to deploy agentic AI within two years (Deloitte). But only 21% have governance ready. That gap is worth paying attention to, both as a risk to manage and as an opportunity for organisations that get their governance sorted early.

Build practical AI skills with structured courses

BUILDING YOUR OWN AI TOOLS

This is where things get interesting for organisations that want to move beyond generic AI use. And where some of the biggest competitive advantages are being built right now.

Scaling elite performance with CustomGPTs covers an idea we're increasingly excited about. When you take a generic AI platform and configure it with your organisation's specific knowledge, processes, and standards, you create something far more useful than the base tool. A CustomGPT that knows your brand voice, your customer segments, your internal processes, your quality standards is orders of magnitude more useful than a generic AI assistant. It moves AI from being a clever novelty to being a genuine operational asset.

And the barrier to building these is dropping fast. You don't need to be a developer. The major platforms now offer no-code or low-code ways to create custom AI tools. Upload your process documents, define the instructions, set the constraints, and you have a purpose-built tool that your team can use immediately. We've seen teams go from idea to working custom tool in an afternoon. That's a capability that was unimaginable two years ago.

AI automation as your assistant coach extends this idea further. When you combine custom AI tools with automation platforms, you can create workflows where AI handles entire multi-step processes. Not just answering questions, but doing work. Invoice processing, report generation, customer onboarding, content review. The list grows as the tools mature. And every workflow you automate well frees your team up for higher-value work.

The key principle is this: generic AI tools are the starting point. Custom AI tools built around your specific workflows are where the real competitive advantage sits. And you don't need a technical team to build them. The opportunity is there for any organisation willing to invest a bit of time in configuring the tools around their own needs.

CHOOSING THE RIGHT TOOL FOR YOUR CONTEXT

We see organisations get stuck on tool selection. They run extensive evaluations, build comparison matrices, form committees. And by the time they've decided, the tools have changed again. The perfect becomes the enemy of the good.

Here's a more practical approach. Start with your workflow, not the tool.

1. Identify your highest-value repetitive workflows. Where does your team spend the most time on tasks that follow a pattern? That's where AI delivers the clearest ROI. Look for the work that's important but repetitive, the kind of tasks where your team knows the process cold but still has to grind through it manually.

2. Match the tool to the ecosystem. Google Workspace? Start with Gemini. Microsoft 365? Start with Copilot. Neither? ChatGPT or Claude, depending on whether you need breadth or depth. This one decision eliminates most of the analysis paralysis, because the ecosystem fit matters more than marginal differences in model capability.

3. Start narrow, then expand. Pick one workflow, one team, one tool. Get good at it. Learn what works and what doesn't. Then expand to adjacent workflows and teams. The organisations that try to roll out AI everywhere at once almost always stall. The ones that start with a focused pilot and iterate almost always succeed.

4. Evaluate the AI, not the demo. The AI dismissal reflex is real. People try an AI tool once, get a mediocre result, and conclude it's overhyped. That's like picking up a guitar for the first time, playing badly, and concluding guitars don't work. Give the tool enough time and iteration to actually show what it can do. Most people are surprised by how much better results get after just a few hours of practice.

5. Budget for training, not just licences. An AI tool without training is like buying the whole team a piano and expecting music. Generative AI's true nature is that it's a capability that develops with practice. The investment in training always pays back more than the investment in the tool itself. And the training doesn't need to be elaborate. Structured practice on real workflows beats classroom theory every time.

Off-the-shelf versus custom AI solutions is a question that comes up a lot. For most organisations, the answer starts with off-the-shelf and moves toward custom as proficiency grows. You need to understand what AI can do before you can specify what you need it to do. And that understanding only comes from hands-on experience.

WHERE TOOLS GO WRONG

Not every AI implementation works. Understanding why helps you avoid the same pitfalls and set up your own deployments for success.

The McDonald's AI drive-through experiment is a case study we reference a lot. They deployed AI into an existing drive-through process without rethinking the process itself. The AI couldn't handle the messy real-world complexity of drive-through orders, accents, background noise, and menu customisations. The lesson isn't that the AI wasn't good enough. It's that bolting AI onto an unchanged process rarely delivers the results you're hoping for. You need to redesign the workflow around what AI can and can't do. The organisations that learn this lesson early save themselves a lot of time and budget.

AI and customer service (Klarna) provides a contrasting example that shows what's possible when you get the approach right. Klarna's AI assistant handled two-thirds of customer service interactions within a month of launch, equivalent to 700 full-time agents. The difference? They redesigned the customer service workflow around AI capabilities rather than just adding AI to the existing one. They thought about the end-to-end experience and built AI into the right parts of it.

The platform evolution itself creates traps. The contrarian view of the AI arms race makes the point that chasing every new model release and every new feature is exhausting and counterproductive. The principles of using AI well stay constant even as the tools evolve. Focus on those principles and you won't be wrong-footed every time a new version ships. Build your team's skills around fundamentals rather than features, and the tool upgrades become a bonus rather than a disruption.

And there's the question of what's useful versus what's marketing. The AI Advantage is about cutting through the hype to find actual business value. Not every AI feature is worth adopting. Not every new model improves your workflow. Developing the judgment to distinguish signal from noise is itself a valuable skill. The teams that develop this judgment early spend their time on AI applications that actually deliver, rather than chasing the latest announcement.

Get weekly insights on the tools that actually matter

THE SKILLS THAT MATTER MORE THAN THE TOOLS

We'll say this directly: the specific AI tool you use matters less than how well you use it.

We've seen people get extraordinary results from ChatGPT Free and mediocre results from enterprise-grade AI deployments. The difference is always skill, not software. And that's encouraging, because skill is something every team can build.

The skills that make AI tools useful are:

Prompting. The ability to give AI clear, specific, contextual instructions. This is a learned skill that improves with practice. A vague prompt gets a vague response. A detailed prompt that includes context, constraints, examples, and a clear definition of what "good" looks like gets dramatically better output. And the skill transfers across every platform. Good prompting on ChatGPT means good prompting on Claude or Gemini.

Iteration. AI rarely nails it on the first try. The people who get the best results treat the first output as a starting point and refine through conversation. "That's close, but make it more concise." "Good structure, but the tone needs to be more direct." "Add specific data points from Q3." This iterative approach is how you move from generic AI output to work product that's actually useful. And it gets faster as you develop an instinct for what direction to push the AI in.

Judgment. Knowing when AI output is good enough, when it needs refinement, and when it's off base. AI doesn't flag its own limitations. It presents everything with equal confidence. The human in the loop needs the domain knowledge and critical thinking to evaluate what they're getting. This is why AI amplifies expertise rather than replacing it. The more you know about your field, the better you can assess what the AI gives you.

Integration. Knowing how to weave AI into existing workflows rather than treating it as a separate activity. The goal isn't to "use AI" in isolation. The goal is to make AI a natural part of how you already work. The teams getting the best results are the ones where AI has become invisible, not a special event, just how things get done.

These skills transfer across platforms. If you can prompt well on ChatGPT, you can prompt well on Claude or Gemini. If you can iterate effectively with one tool, you can iterate with any of them. Invest in the skills and the tool choice becomes secondary. That's the highest-return investment you can make in AI right now.

FURTHER READING FROM ACROSS THE SITE

FURTHER READING FROM ACROSS THE SITE

Major Platforms

• Going all in on Gemini and Google - Why we chose the Google ecosystem
• Microsoft Copilot is now world class - The enterprise AI option
• Gemini Advanced arrives - Google's premium AI tier
• Copilot Pro reshaping business landscapes - The Microsoft AI journey

AI Agents and Automation

• AI agents: what they really are - Cutting through the agent hype
• AI agents are coming - What agent technology means for business
• What sets you apart when agents are everywhere? - The human differentiator
• AI automation as your assistant coach - Practical automation approaches

Practical Skills and Use Cases

• Scaling elite performance with CustomGPTs - Building your own AI tools
• AI just got Excel superpowers - AI for data work
• The AI dismissal reflex - Why people give up on AI too early
• Generative AI's true nature - Five use cases beyond content generation
• AI rewards a questioning mindset - The skills that matter for AI proficiency

Case Studies and Cautionary Tales

• McDonald's failed AI drive-through - What happens when you bolt AI onto unchanged processes
• AI and customer service (Klarna) - How Klarna redesigned customer service around AI
• AI is a threat to search advertising - How AI tools are disrupting Google's business model
• Off-the-shelf vs custom AI solutions - Build versus buy considerations

ABOUT THE AUTHOR

Mark Laurence

Mark is the founder of Ten Past Tomorrow, an AI consultancy and education business based in New Zealand. A trained futurist (Institute for the Future) and practical AI specialist, he works with senior leadership teams to move organisations from AI curiosity to AI capability.

He has worked with 100+ NZ organisations and leads Rapid AI Traction, a four-week programme for senior leadership teams, and The Path to AI Emergence, a ten-month transformation programme.

EXPLORE MORE GUIDES

BUILD YOUR AI SKILLS

or sign up to our newsletter