AI tools for business: start with what you need them for, then pick the tool

Key Takeaways

• The AI tool is the part that lasts the least. You use one today, and in three or six months a better one appears or the one you had starts doing the work of several. What lasts is having clear use cases: which specific task in your business you want AI to solve.
• In 2026 alone, NotebookLM was renamed Gemini Notebook and Microsoft 365 Copilot started being renamed Microsoft Copilot. A list of "the best tools" goes stale in weeks.
• In Colombia, the most common uses of AI at work are searching for information (55%), drafting emails (30%), and summarizing documents (28%), and surveyed workers report saving 8 hours a week on average (EY, Work Reimagined, February 2026).
• Many companies already pay for AI without knowing it. Gemini is included in Google Workspace Business plans, and Copilot Chat in eligible Microsoft 365 licenses.
• The free and paid versions are different tools. Frontier models are in the paid plans, and for privacy an individual paid plan isn't enough: only the business plans of ChatGPT, Claude, and Gemini stop training on your data by default.

When a manager asks us which AI tool they should buy, we almost always answer with another question: what do you want it for? It's not a dodge. The tool we recommend today probably won't be the one they use a year from now, and if they don't understand what the technology is for, they'll choose by advertising.

This guide does name tools, but it organizes them by what they do, not by brand. It explains the AI capabilities worth understanding, which tools offer them today, what they cost, what they do with your data, and how to choose without getting locked into any of them. It's part of our guide to artificial intelligence for businesses.

Why is the AI tool the least important part?

The tool is the least important part because it changes faster than a company can adopt it. In 2026 Google renamed NotebookLM as Gemini Notebook, Microsoft started renaming Microsoft 365 Copilot as Microsoft Copilot, and in the Gemini app the Veo video model is being replaced by Gemini Omni. That's before counting the new tools that appear every month.

What doesn't change as fast are the capabilities. Two years ago AI could already draft, summarize, search, and interpret documents, and it still does, only better and cheaper. If your team understands that it needs "a tool that reads purchase orders and extracts the data," it can change vendors without changing the process. If it only learned to use one product, every change forces it to start over.

That's why the key word isn't tool, it's use case: a specific task, with an owner and a measurable result, like "classifying the requests that arrive in the customer service inbox." Once the use case is clear, the tool almost identifies itself, and when the tool changes, the use case is still there. In the programs we run we don't teach tools, we teach people to find use cases and solve them with AI's capabilities. The tool is chosen last, as we explain in how to implement artificial intelligence in a business.

Which AI capabilities are useful to a business?

A mid-sized company can benefit from eight AI capabilities, and each one solves a different kind of use case. The tools in the third column are examples as of September 2026, not permanent recommendations.

The first two rows cover what people use most today. According to EY's Work Reimagined study, surveyed Colombian workers use AI mainly to search for information, draft emails, and summarize documents. The automating and building rows are the ones that change processes, and that's where the biggest gains are. To see when each one makes sense, read business process automation: when you need AI and when you don't.

Which AI assistant should you choose: ChatGPT, Claude, Gemini, or Copilot?

The best starting point is usually the assistant that comes with what your company already pays for. If you work with Google Workspace, Gemini is already included. If you work with Microsoft 365, Copilot Chat is already included in eligible licenses. Adding another assistant makes sense when the team already uses it heavily and needs something the included one doesn't do.

Prices change often and some are shown in local currency depending on where you view them. Treat them as an order of magnitude and check the current price before buying.

Why isn't the free version enough?

It isn't enough because the free version of an AI assistant is a different tool: it has less capable models, lower usage limits, and different privacy rules. Most people got used to using free versions for everything, including work with company information, and that has a cost in quality and in risk.

In capabilities, frontier models, the most capable ones each vendor offers, are in the paid plans. The names change every few months, so this table is a snapshot from September 2026:

The difference shows in the tasks that matter: analyzing a long document, reasoning about numbers, or writing a proposal that doesn't sound generic. Judging what AI can do for your company from the free version is like evaluating a cargo truck by driving the showroom model.

In privacy, the nuance almost nobody knows is that paying isn't enough. Individual paid plans, like ChatGPT Plus or Pro, Claude Pro or Max, and Google AI Pro, are consumer plans and follow the same rules as the free version. Only business plans change the default rules.

The sources are the policies of OpenAI, Anthropic's consumer terms update, and the Gemini privacy hub, which asks users not to enter confidential information they wouldn't want a reviewer to see. If someone on your team uses a personal account, the minimum is to turn off the training setting: "Improve the model for everyone" in ChatGPT, the training setting in Claude's privacy options, and Gemini Apps activity. The right answer is to give them a business account.

What is an IDE and why does it now matter to a manager?

An IDE (integrated development environment) is, in AWS's definition, a software application that helps programmers develop code efficiently, bringing editing, running, and testing software together in one place. Until recently it was a tool only programmers used.

That changed because IDEs now come with AI agents that write code from natural language instructions. Google Antigravity, launched in November 2025, is an IDE designed from the start to work with agents that plan and carry out tasks. Tools like Claude Code and OpenAI's Codex do something similar from the terminal or the browser, and platforms like Replit or Google AI Studio let you build and publish apps without installing anything.

For a manager, the practical consequence is that the person who knows the process can build the first version of the tool they need, with support, instead of writing a requirements document and waiting for a vendor. In El Salvador, eight advisors with no programming background built working web systems, one of them a five-module management system in 17 hours of mentoring.

A warning: on some of these platforms the free tier does use your content to improve their products. Google AI Studio says so explicitly on its pricing page. Don't use free tiers with customer information. We explain this in AI acceptable use policy.

How to choose AI tools for your business, step by step

This is the order we use with companies. It works for choosing the first tool and for deciding, six months later, whether to switch.

1. Define the use case, not the tool

Write down the use case: the specific task you want to improve, who does it, and what it costs today. "Each quote takes the sales team three hours to prepare" is a use case, "we want to use AI" isn't. The use case defines everything else, and a list of five or six well-written use cases is worth more than any tool comparison.

2. Identify which capability the use case needs

Use the capabilities table. A quote needs drafting and data matching; a competitor report needs deep research; loading invoices needs reading documents and automation. Many tasks combine two capabilities.

3. Check what you already pay for

Before buying, check what your office suite, ERP, or CRM includes. Many companies discover they already have an included assistant nobody switched on or the team doesn't know how to use.

4. Check what the tool does with your data

Look at the vendor's privacy page to see whether it uses your information to train models, where it stores it, and how it's deleted. If the tool will touch customer data, require a business plan with a corporate account.

5. Test it for two weeks on a real task

A demo tells you nothing. Have two or three people use the tool on the task from step 1, with real, permitted information, and compare against what it cost before.

6. Build so you can switch tools

Keep your instructions, templates, and data in company documents, not inside a single platform. If you build your own tools, keep the code in a company repository. That way, when something better appears, you change tools without losing the work.

Real case: the tools we use at Suricata Labs

At Suricata Labs we don't have a favorite tool. We have capabilities we use every day and tools that rotate. For deep research we use the Deep Research features of Gemini, Claude, and ChatGPT, and mix them depending on the task. We use ChatGPT for text and also for images, and Gemini for images, video, and text. For working on our own documents we use Gemini Notebook, and Gemini's AI inside Google Drive is, in practice, what the team uses most day to day.

Where we get the most value is building. We use Claude Code and Codex inside an IDE, currently Antigravity, although which IDE we use makes little difference to the result. Sometimes we use Replit and Google AI Studio for quick prototypes. With those tools we built, among other things, the agent that researches companies: we give it a company's name or website, and in 5 to 10 minutes it delivers a report that used to take a junior consultant about 10 hours. We tell that story in what an AI agent is.

For analysis that needs a decision rather than text, we use TypeSafe's Jev, a model designed for software to make decisions. Instead of writing, it answers specific questions with a structured answer and a confidence level: it picks an option from a list, assigns a score, or says whether something is true. Today we use it to classify information, verify data, and prioritize, and we keep experimenting with other use cases. It's a good example of why the use case comes first: with the task clear, the right tool wasn't a chat assistant but a specialized model that plugs into our own tools.

What we use most today, besides what comes in Drive, are the tools we've built ourselves. And the list in this section will probably be different six months from now. What won't change is the question we use to choose each one: which use case is it for, and which capability does it need.

Frequently asked questions about AI tools for business

What is the best AI tool for business?

There's no single best AI tool for every business, because it depends on the use case and on what the company already uses. For most, the best starting point is the assistant included in their office suite: Gemini if they work with Google Workspace, or Copilot if they work with Microsoft 365. Specific tools are then added for the tasks that assistant doesn't cover.

Are there free AI tools for business?

Yes, several tools have free tiers, but it's best to use them only with public or test information. Free versions use less capable models and, depending on their settings, may use your conversations for training, just like individual paid plans. To work with company or customer information, a business plan with a corporate account is the safer choice.

Is ChatGPT, Gemini, or Claude better?

All three handle a company's most common tasks well, like drafting, summarizing, and research, and they overtake each other every few months. The practical decision depends on what you already pay for, what each does with your data, and which works best on your specific task. At Suricata Labs we use all three and mix them depending on the task.

Do I need to know how to code to build tools with AI?

Not to get started. Today's platforms let you build internal apps by describing what you need in natural language. In our programs in El Salvador, advisors with no programming background built working web systems. It does help to have support the first few times and clear rules about which data can be used.

How much do AI tools cost for a business?

The best-known business assistants cost around USD 20 per user per month billed annually, and in many cases they're already included in the office suite the company pays for. Automation and building tools have similar entry plans. The cost that weighs most isn't the license, but the team's time to define the task and learn to use the tool well.

The question that outlasts any tool

The tools in this guide will change. Some will disappear, others will be renamed, and others will absorb what three different ones do today. What helps your company is having clear use cases and knowing which capability each one needs, because with that answer any new tool can be evaluated in an afternoon.

If you want to identify which AI capabilities help your company and build the first tool with your team, at Suricata Labs we do it with the people who know the process.

Explore our AI services | Schedule a conversation

Read also: What is an AI agent and what is it for in a business