The number of AI tools has grown quickly and continues to grow. Evaluating all of them is not a practical use of anyone’s time. A more useful approach is to start with the problem to be solved and then find a tool that fits it.
Start with the task
Three things need settling before looking at any product:
- The specific task the tool is meant to make easier or faster, described precisely enough to test. “Content creation” is too broad, and “writing a first draft of the monthly newsletter” is specific enough to evaluate against.
- How long the task takes now, because a task that takes ten minutes a week rarely justifies the setup time, and one that takes half a day deserves a proper look.
- What good output looks like, because knowing the target makes it much easier to judge whether a tool is delivering it.
The main tools and what they suit
Most organisations find that a small number of platforms cover what they need. The major general-purpose tools, ChatGPT, Claude, Gemini and Copilot, all handle writing, research and summarising well. The differences between them matter less than how well each one is configured and prompted.
An organisation already working in Google Workspace or Microsoft 365 may find the matching assistant easier to adopt, and our FAQ on Google Workspace covers what that platform includes.
Beyond the general-purpose tools there are specialist tools for image generation, video, audio, coding, data analysis and specific applications such as customer support or scheduling. These can be explored once the basics are working.
Check what happens to the data
Before committing to a tool, check where the information entered into it is stored, whether it is used to train the model, who in the organisation can see it and how the terms differ between free and paid plans. These terms change over time and need to be read for the specific plan in use. Confidential and personal information should stay out of a tool until those terms are understood. Rules for staff use are covered under Technology Governance.
What to avoid
- Paying for several tools that do the same thing, when one well-configured general-purpose tool usually covers most needs.
- Choosing a tool because it is getting the most media coverage this month, since the tool that is trending may not suit the tasks at hand.
- Assuming the most expensive option is the best one, when a trial on real tasks shows far more than a price list.
Get the foundations right first
AI tools work best on reasonably clean data, clear processes and a digital setup that works. If files are disorganised and basic tools are not working properly, AI adds another layer to an already messy setup. Digital Setup is a better starting point than any AI tool in that situation.
How we help
Our AI Automation service includes vendor-neutral tool selection, so recommendations are based on the organisation’s needs, security requirements and scale. It also includes a short proof of concept, which tests a tool on the organisation’s own material before budget is committed.
When a tool has been tried and the results were underwhelming, the cause is usually configuration. Setting the tool up with the organisation’s voice, audience and terminology makes a large difference, and our FAQ on getting AI content to sound like the organisation covers that question.
