AI 101 Aware
Get More from Every Token
3 July 2026
Use AI more efficiently by reducing unnecessary token usage while maintaining high-quality results.
Why this matters
Every interaction with an AI model consumes tokens - the units used to process both your prompt and the AI's response. For enterprise AI platforms and APIs, token usage affects cost, response speed and how much context a model can retain.
Using AI efficiently is not about writing shorter prompts. It is about providing the right information at the right time. These six habits help reduce costs and improves response quality and get more value from every interaction.
1) Choose the right model
Not every task requires the most capable reasoning model.
Simple tasks such as summarising emails, rewriting text or extracting information often perform well using smaller, faster models. Reserve more advanced reasoning models for activities such as analysing policy options, writing complex code or evaluating multiple scenarios.
Using the appropriate model can significantly reduce costs while maintaining quality.
2) Use Projects to keep context
One of the biggest sources of wasted tokens is repeatedly providing the same background information.
Instead of copying your instructions into every new conversation, use Projects (where available) to store writing style, background information and standard instructions.
A Project in ChatGPT
3) Retrieve only what you need
Only provide documents that are relevant to the task.
When using connectors such as SharePoint, ask AI to search for specific files rather than your entire Sharepoint. This keeps responses focused and avoids processing unnecessary information.
4) Start focused
The clearer your request, the fewer follow-up questions the AI needs to ask. A useful structure is to include three pieces of information up front:
5) Keep the conversation going
Continue related work in the same chat so the AI can reuse existing context.
6) Save and reuse prompts
Save prompts you use regularly for tasks such as:
Summarising reports
Submission writing
Email drafting
Reusable prompts improve consistency and reduce repetitive prompting.
Key takeaways:
Getting more value from every token is about using context deliberately, not writing the shortest prompt.
Choose the right model, avoid repeating information, retrieve only relevant content and reuse successful prompts. These simple habits reduce unnecessary AI processing while improving the quality and consistency of your results.
↩️Return to: All Guides & Resources