Decide What to Automate and What to Leave Alone

    You are an automation strategist who has seen a lot of automation projects get cancelled. Help me decide what is worth automating.

Tasks I'm considering: {{list_of_tasks}}
Who does them now: {{who}}
What tools I have available: {{platforms_scripts_ai_tools}}
My appetite for build effort: {{time_and_budget}}

Score each task on six dimensions, 1-5, and show your reasoning per score:

1. Volume — how often it runs. Low volume rarely repays the build, however annoying the task.
2. Time per run — including the context switch to start it, which people forget to count.
3. Stability — has the process changed in the last year? An unstable process automates into a maintenance burden, and every change now costs engineering time instead of a conversation.
4. Rule-clarity — can the decision points be written as explicit rules? If the answer is "you just know," it is not ready. Write the SOP first; automation is the second step, never the first.
5. Failure cost — what happens if it runs wrong at 3am and nobody notices for a week. This is the dimension people skip, and it is the one that kills projects.
6. Input quality — is the input structured and reliable, or is it whatever a human typed into a free-text field?

Then classify each task:
- Automate now: high volume, stable, clear rules, low failure cost
- Document first: worth automating, but the process is not specified well enough yet
- Assist, don't automate: let a human stay in the loop on the judgement, automate the fetching and formatting around it
- Leave alone: low volume, unstable, or high consequence
- Delete instead: the real answer for some tasks — they should not be done at all, and automating them just makes the pointless work faster

For anything in "automate now," give me:
- The simplest thing that could work, before the sophisticated thing
- What has to be monitored, and what alerts when it fails — an automation nobody watches is a liability, not an asset
- The manual fallback when it breaks, and who knows how to run it
- The payback period in hours saved versus hours to build and maintain

Be honest about maintenance. A rough annual maintenance cost of 20-30% of the build effort is a reasonable planning assumption; say when you think it is higher.