Not every Human/AI skill requires a workflow, a maturity framework, and a structured handoff. Some are just habits. Good ones, but habits.
Daily Augmentations are the foundational layer of Human/AI collaboration. They are the tasks where AI assistance is straightforward, immediately useful, and accessible to anyone regardless of their technical background. They do not require deep prompt engineering or an understanding of model architecture. They require only the willingness to try.
The value of getting these right is not the individual task. It is what they free up. Every hour recovered from low-leverage work is an hour available for the thinking that only you can do.
The Tasks
| Task | What to ask | Recommended model tier | Key watch-out |
|---|---|---|---|
| Email cleanup | "Rewrite this to be clearer and shorter. Keep the tone." | L1 — any capable model | Read the output. Models sometimes cut content that matters. |
| Meeting notes | "Here are my rough notes. Give me a structured summary with action items and owners." | L1 — any capable model | Verify action items against your memory of the meeting. Models infer; they do not attend. |
| Calendar and scheduling | "I have these constraints. Suggest a schedule for the week that protects focus time." | L1 — any capable model | You know your energy patterns. The model does not. Treat the output as a first draft. |
| Basic code generation | "Write a [language] function that does [specific thing]. Include error handling." | L2 — a reasoning-capable model | Always read generated code. Never deploy without understanding what it does. |
| Quick research synthesis | "Summarise the main arguments for and against [topic] in plain language." | L2 — a reasoning-capable model | Verify any specific claims, statistics, or attributions before using them. |
| Document formatting | "Restructure this into [format: bullet points / numbered list / table]. Preserve all information." | L1 — any capable model | Check that nothing was dropped. Formatting requests can quietly remove nuance. |
| First draft of routine comms | "Draft a [thank you note / project update / brief status report] based on these facts." | L1 — any capable model | Add your voice. Routine comms drafted by AI are fine. Routine comms that read like AI are noticed. |
Human Maturity
SFIA level 3 minimum, but accessible to anyone who uses email. The limiting factor for Daily Augmentations is not skill. It is habit. The practitioners who benefit most are those who build a consistent practice of reaching for AI assistance on low-value tasks before reaching for their own attention.
The most common failure mode is not misuse. It is under-use. The majority of knowledge workers who could save four to six hours per week from Daily Augmentations are not doing so. They have the tool. They have not formed the reflex.
Model Maturity
L1 to L2. Most Daily Augmentation tasks are within the capability of any current frontier model at standard settings. The key distinction is between tasks that require accuracy (code generation, research synthesis) and tasks that require only adequacy (email cleanup, document formatting). Accuracy tasks benefit from a reasoning-capable model. Adequacy tasks do not.
A note on the Key Watch-Outs column
Every row in the task table has a watch-out. This is intentional. Daily Augmentations are safe at scale precisely because they keep the human in the loop. The watch-outs are the loop.
AI does not know your meeting as well as you do. It does not know which email recipient will read a shortened version as dismissive. It does not know that the statistics it found are from a source with a conflict of interest. You do. That knowing is the human contribution that makes the augmentation worth doing.
Business Area Impact
Daily Augmentations, when adopted consistently across a team or organisation, change the workload profile in ways that are invisible individually but significant in aggregate.
Administrative support functions feel the effect first. Tasks like scheduling, note-taking, and routine correspondence formatting represent a substantial proportion of administrative roles. When those tasks are handled directly by subject matter experts with AI assistance, the administrative function either shrinks or shifts toward genuinely complex coordination work.
The roles least at risk are those whose administrative work is inseparable from relationship management and contextual judgment. A personal assistant who knows which meeting is genuinely urgent and which can wait is not doing the same job as an AI scheduling assistant. The risk is to the parts of the role that AI can handle, not the whole.
The organisations that navigate this well do so by helping people identify which parts of their role are augmentable and which are not, and by treating the time recovered as an investment in the non-augmentable parts rather than a headcount efficiency gain.
Handoff
Daily Augmentations are mostly self-contained. The output of email cleanup is a sent email. The output of meeting notes is a distributed summary. The output of code generation is a working function.
Where Daily Augmentations feed into a larger chain, the handoff is informal: the email goes out, the notes land in the project space, the code gets committed. The discipline is not in the handoff protocol but in the output review step that happens before it. Check before you send. The model worked quickly. That is the point. It does not always mean the output is right.