Services
Explore
Start an AI assessmentAnswer 15 questions about one process and see where AI could help.
Start an assessmentWhat is generative AI, and what could it mean for your business?
How generative AI works, what it is good at, where it goes wrong and how to decide whether it suits a task in your business.
The short answer
What AI and generative AI mean
Generative AI produces text, images, audio, code and other content using patterns learned during training. Your instructions and any material you provide shape the output. It can produce useful drafts, and it can also invent details or get things wrong. ChatGPT, Microsoft Copilot and Claude are examples of tools built on it.
Artificial intelligence is a broad term for software that carries out tasks such as understanding language, recognising patterns, making predictions and producing content. It covers many techniques and sits underneath products such as ChatGPT, Microsoft Copilot and Google Gemini.
How it works
A simple way to think about how it works
Generative AI models are trained on very large amounts of text and other data. From that training they learn patterns: which words tend to follow others, which ideas tend to appear together and what kinds of answer tend to follow what kinds of question.
When you give a model a prompt, such as a question, an instruction or some background material, it uses those patterns to produce a likely response. It does not check facts the way a person would. Its output depends on the patterns it learned, the instructions it is given and any material you supply.
Realistic expectations
What generative AI can and cannot do
Can do
Draft written content such as emails, reports, summaries and descriptions
Answer questions from documents or reference material you provide
Pick out key points from long, unstructured text
Produce images, audio or video from written instructions
Hold a conversation and follow multi-step instructions
Write and review code, spreadsheet formulas and structured data
Cannot do reliably
Guarantee factual accuracy: it can state wrong information confidently
Know about recent events unless it is connected to a current data source
Take responsibility for high-stakes decisions, which stay with a person
Use your business information unless it is configured with your documents or connected to your systems
Understand your business context unless you describe it
Business applications
Where businesses use generative AI
These are common examples by business function. Whether each one suits your business depends on your volume of work, your data and the checking involved.
- Drafting blog posts, social captions and email campaigns from a brief
- Adapting existing content for different audiences or channels
- Producing product descriptions in volume, for a person to check
- Summarising long reports, contracts or meeting notes
- Extracting data from documents for entry into other systems
- Answering staff questions from your internal documents
- Drafting personalised outreach for a salesperson to review
- Preparing proposals and quotes from a template
- Sorting incoming enquiries before they reach your team
- Answering routine enquiries outside office hours
- Drafting responses to complaints for a person to review
- Sorting and routing support tickets
Tools and plans
Using AI tools compared with having a plan for AI
Many businesses start with individual tools: someone begins using ChatGPT, or a few Copilot licences are bought. That is a reasonable way to learn what the tools can do.
A plan for AI sets out
tasks in your business could benefit,
tools or custom work suit them,
your team will use them consistently, and
Common questions
Questions about generative AI
Is generative AI the same as ChatGPT?
ChatGPT is one product built on generative AI, using large language models made by OpenAI. Generative AI is the wider category. Other examples include Google Gemini, Anthropic's Claude and Meta's Llama.
What's the difference between AI and automation?
Traditional automation follows fixed rules: if X happens, do Y. It is predictable and works well when the input is consistent. Generative AI can work with varied, unstructured input such as free-text emails, but its output is less predictable and needs checking. Many tasks are best handled by ordinary automation, and some combine both.
Do I need to share my business data with an AI company?
Not always. Some setups keep data within your own systems or a private cloud environment, and providers offer plans with different data terms. The right setup depends on the data involved and the risks you are willing to accept. Check the provider's current terms for the plan you use.
Can generative AI make mistakes?
Yes. AI models can produce confident answers that are wrong, which is often called hallucination. Check anything that matters before it is used. For many business tasks, AI produces a first draft that a person then checks and corrects.
How is generative AI different from the AI that's been around for years?
Many long-standing uses of AI focus on recognition and prediction, such as spam filters, recommendation engines and fraud detection. Generative AI produces new content, such as text, images, code and audio. That makes it useful for a different set of tasks, particularly those involving writing and conversation.
How do I know if my business is ready to use AI?
Start with a task you want to improve that happens often and is easy to check. If you can describe the task, the information it uses and who would check the output, you can test it. An AI discovery workshop is one way to find and compare candidate tasks.
Talk to us
Want help applying this to your business?
Bring the task that takes your team the most time. We reply within one working day.