The AI+ Library — byStraits Institute for Applied AI
AI+Libraryby Straits Institute for Applied AI
You
Issue no. 07 · 30 September 2026
Issue no. 0730 September 2026

Which of your tasks will AI take first?

Your Essential Report from the Frontlines of Applied AI

Editor's note

Spreadsheets didn’t eliminate accountants. They eliminated manual calculation.

The same is happening with AI now, one task at a time.

England’s planning officers spent an estimated 250,000 hours a year checking decades-old planning records by hand. Since June, an AI tool called Extract turns those records into usable data in minutes, at every council in England.

The planning officers are still there. According to Ian Murray, the UK’s Minister for Data and Modern Digital Government, the tool is “taking admin off their desks so they can focus on the skilled work their communities need most.”

So which of your tasks will go?

A job is just a bundle of tasks. Many of these tasks will be completed differently with AI.

It’s clear that the tasks will be affected in a predictable order:

  • Searching, checking and re-keying go first.
  • Judgement, and dealing with people, become more important.

The people most at risk are the ones who keep doing the tasks in the first category.

Your job stays. But it will change shape. So know which of your tasks are which, and move your time towards the ones that grow.

The people who move first get to choose which tasks they keep.

So this week is about sorting your own tasks. There’s a prompt below from AI+ Mental Wellness that sorts your recurring tasks into what to hand over, what to speed up and what to protect, and a case study on England’s planning officers.

Plus the week’s news in applied AI — land officers in Taiwan, CCTV searches in Seoul, freight-wagon checks in India and delivery drivers in Brazil. In each one, a task moved to AI and the person stayed.

Key idea

Every job is a bundle of tasks. Tasks that can be completed differently with AI.

So how do you find out which of your tasks will change? Start with what you actually do in a week.

Write down the tasks behind your job title. A librarian’s week, for example, includes chasing overdue loans, cataloguing new books, answering research enquiries and helping students judge which sources to trust.

Then sort each task into one of three groups:
AI can do it. Chasing overdue loans. Hand it over, and check what comes back.
AI can speed it up. Cataloguing, and first answers to enquiries. You still do the work; AI gives you the first draft.
Only you can do it. Helping a student judge a source. It matters more now, so this is where the time you save should go.

Not sure where a task belongs? Try AI on it once, and check the result before you decide.

Then move your time. Hand off the first group, speed up the second, and put what you save into the third.

The third group is the part of your job that grows.

AI+ Everyone
Source
AI+ Everyone
Chapter 9 — how AI changes your job, one task at a time.
Latest News in Applied AI
  • Public administration

    Tainan’s land office lets AI screen property-price filings, and officers check what it flags

    The AI has checked 142,758 price declarations against local market prices and flagged 2,558. Land officers checked those and found 74 that needed correcting, often a price typed with a digit too many or too few. The bureau’s words: AI raises the warning, the professionals make the judgement.

    United Daily News, 26 September 2026

  • Public safety

    Seoul’s Gwangjin district uses AI to search CCTV when someone goes missing

    Given a photo or description of a missing child or older person with dementia, the system searches an hour of footage from 100 cameras in about ten minutes. Staff used to check each camera’s footage one by one. It maps the person’s movements over time so police can set the search area.

    Web Economy, 24 September 2026

  • Railways

    Indian railway cameras now check the freight-wagon seals staff used to inspect by hand

    Northeast Frontier Railway and IIT Guwahati launched Project DRISHTI on 25 September. Cameras at checkpoints read wagon numbers, check doors, locks and seals for tampering, and record what they find for staff to act on. The general manager said hand checks were hard at night and in bad weather.

    The Sentinel, 27 September 2026

  • Logistics

    Seara’s delivery drivers report to an AI assistant by WhatsApp voice note

    Seara, part of JBS, has begun rolling out Mari, an AI assistant built by uMov.me, to 300 drivers across 20 distribution centres. Drivers tell it about late arrivals or part-deliveries by voice note, with no new app to install. When a route goes off plan, it alerts supervisors and the back office.

    Agrimídia, 23 September 2026

Do this faster with AI
The task
Sort your week’s tasks into what AI can do, what it can speed up, and what to keep

Paste the tasks that fill your week and it sorts each one into four groups: hand to AI, speed up with AI, keep for yourself, or try AI on first. It ends with the three tasks to hand over first. Keep the task descriptions general — no client names, colleagues’ details or figures.

I'm going to paste a list of the recurring tasks that fill my week as a [your role]. For each one, sort it into exactly one of four categories and explain your reasoning in one line:
- Automate: repetitive, rule-based, low-judgement — could be handed to AI or a simple automated workflow
- Augment: needs my judgement, but AI could speed up a first draft or the thinking
- Protect: high-judgement, relational, or meaningful — should stay fully human and be defended
- Explore: might benefit from AI but needs a small test first
At the end, list the three tasks where handing work to AI would free up the most time and mental energy for me. Here are my tasks: [paste your list].

Prompt 8.1 from AI+ Mental Wellness, word for word.

AI+ Mental Wellness
Source
AI+ Mental Wellness
Chapter 8 — burnout at work, and moving the draining tasks off your plate.
Best practices in Applied AI
Spotlighted practitioner
Planning officers, England

Take the record-digging off the planning officer’s desk, and keep every decision with the officer.

England’s planning officers spend an estimated 250,000 hours a year checking old planning documents by hand — decades-old papers and maps, some with handwritten notes.

Extract, an AI tool from the Ministry of Housing, Communities and Local Government, turns those records into usable data in minutes. After trials with 20 councils, including Exeter and Hillingdon, it has been available to every council in England since 17 June. The government expects it to cut the average council’s manual work on these documents from more than 500 hours to around 255.

A second tool, in testing with Barnet, Camden and Dorset, triages householder applications and gives the officer an initial assessment. The aim is to halve decision times from eight weeks to four. Every assessment is reviewed and approved by a qualified planning officer before any decision is made.

If your job runs on old records, the searching and re-keying is the task to hand over first. The decision stays with you.

AI+ Municipal Services Managers
Source
AI+ Municipal Services Managers
Chapter 1 — what AI is already doing in local government, with the sources.
That was issue no. 07 — here is what the next one does for you

Useful tips and tricks for using AI in your job.

Every week: an idea you can act on, a prompt you can use today, the developments that matter, and how another organisation made it work.

Weekly. Unsubscribe in one click. We don’t sell or share your address.

  • An idea you can act on

    One argument, properly made, that changes how you approach your own work.

  • A prompt you can use today

    Paste it straight into your work — taken word for word from the books.

  • Updates on actual AI deployments globally

    What organisations actually put into production, each cited to its source.

  • How another organisation did it

    A real deployment, what it took, and the numbers behind it.