“AI capabilities are advancing rapidly: what began as a chatbot that answers questions is evolving into a work tool capable of using source files, planning work steps, and producing outputs that are ready for review. At a university, it can support a wide range of activities—from preparing teaching materials and research and analysing data to project management, document creation, support services, and software development. The greatest benefit comes when AI helps reduce repetitive manual work, leaving people more time to focus on substantive decisions.
For example, an employee can add one or more Excel files to ChatGPT Work and ask in plain language which metrics have changed, where inconsistencies occur, or what should be examined more closely. Work can compare the data, create a summary and charts, and return the results as a new Excel file. Instead of writing complex formulas, users can simply start by describing their question or objective.
ChatGPT Edu’s capabilities have expanded significantly in recent months. In addition to regular conversations, the same environment can now be used to give AI more comprehensive tasks, bring together materials on a broader topic in a project, use information available in TalTech’s work tools, and create specific outputs—for example, documents, presentations, spreadsheets, analyses, and web applications,” said Raul Hanson, Head of TalTech’s IT Department.
When to choose Chat, Work, or Codex?
| Mode | What is it suitable for? | Examples |
| Chat | Questions, discussion, and shorter tasks | Asking for explanations, developing ideas, improving text, searching the web for information |
| ChatGPT Work | Completing a larger task with a clear final deliverable from start to finish | A report, presentation, comparison table, project plan, recurring overview, or another finished file |
| Codex | Technical work involving local files, data, and code | File processing, data analysis, creating a website or application, modifying and testing code |
A simple rule of thumb is: if you need an answer or want to discuss a topic, start with Chat; if you need a finished deliverable, choose Work; if the task involves technical work or working directly with a project folder on your computer, use Codex.
Default settings are suitable for most tasks
The default settings are suitable for most everyday tasks. There is no need to change the model, reasoning level, or speed unless you have a specific reason to do so.
Standard use of ChatGPT Edu remains included in the licence. In addition, each user receives 800 credits per month for more resource-intensive features.
You can monitor your usage under Settings → Usage. Credits should be used thoughtfully: choose an option that is sufficient for the task while using fewer resources whenever possible.
More context and tools for your work
Conversations, files, and instructions related to a longer-term topic can be kept together in a project.
With your permission, plugins can also allow ChatGPT to access information available to you in tools such as Outlook, Teams, and SharePoint.
How to get started?
The easiest way to get started is with the web version of ChatGPT at chatgpt.com.
- Choose one small, real work task. Start, for example, with a short overview, comparison table, or presentation where the result is easy to verify.
- Describe the desired outcome. Specify who it is for, the format and length you need, and which sources should be used.
- Provide only the necessary context. Start with a small number of source files and clearly state what ChatGPT should not assume.
- Review the result and refine it within the same task. Ask for specific changes instead of unnecessarily restarting the entire task.
If the same workflow starts to recur, it can later be turned into project-level instructions or a Skill. Local Work or Codex is worth trying when the web version is no longer sufficient and the task needs to work directly with files or applications on your computer.
Always review the result
ChatGPT can help complete the work, but a person remains responsible for the final result. In particular, facts, calculations, sources, and personal data must be checked, and the TalTech Rules for the Use of Artificial Intelligence must be followed.