Skip to the content
Projects & Experiments

Job Applications. A graduate job-hunt app, built with AI in an evening.

A private web app that runs one graduate’s job hunt: about 260 researched graduate schemes, every application and where it’s got to, tests and interviews, careers fairs, CVs, ready-made answers for application forms, reminder emails and a calendar feed. Mark Pollak built it for a graduate in the family on the evening of 9 October 2026, with Claude, Anthropic’s AI, writing every line of the code. It was live, with all the research loaded, 56 minutes after the first prompt.

Project
Personal, for a graduate in the family
Built
9 and 10 October 2026
Time
About 4½ hours
Built with
Claude Code, on Claude Opus 5.5
Code typed by hand
None
The Job Applications dashboard: a This week panel with an overdue next action in a red box at the top, then tomorrow’s coding test and online tests, Monday’s interview preparation and Tuesday’s careers fair and technical interview, each with coloured status, tier and listing labels
The same dashboard on a phone, the overdue item first in red

The dashboard

Everything due in the next seven days, anything overdue first in red, then what’s closing soon. Shown here with made-up employers.

  • Web app
  • Phone
  • 56 minFrom the first prompt to a live app, with all the research imported
  • 4½ hWorking time in total, over an evening and the next morning
  • 234Automated tests written alongside the app, all passing
  • 0Lines of code typed by hand
01

A spreadsheet can’t run a job hunt

Applying for graduate schemes is a project in itself. Most open between October and December, each has its own deadline, form, online tests and interviews, and some close early when they’ve had enough applicants.

The research for this job hunt produced a big Google Sheet: about 200 graduate programmes in the UK and about 50 abroad, a deadline calendar, careers fairs, PhD options and hundreds of links to the pages each fact came from. A sheet is good for research and bad for running a job hunt. It can’t tell you what’s due this week, keep a history of each application, hold the CV that was sent, remind anyone of anything, or show which listings need checking again.

So the brief for the app started with one test:

…opens the app and within ten seconds can see what’s due this week and where every application stands.

The brief, 9 October 2026
02

Research first, then a plan

Before any code, Mark asked Claude to research the UK graduate market for the graduate’s degree and interests, and to come up with a strategy. That afternoon it produced:

  • A nine-page plan: who to apply to, in A, B and C lists, how many applications to aim for, and a week-by-week timeline to September 2027.
  • A 76-page report on every scheme: deadline, salary, eligibility, the selection process and the page each fact came from, with a whole section on what it couldn’t verify.
  • A tracker spreadsheet: 200 UK programmes, a deadline calendar, events, PhD options and 255 sources.
  • A second pass that only checks: the 16 most important deadlines, re-read on the employers’ own pages.
  • Three versions of the CV, one for each kind of role, and a cover letter template with answers to common form questions.

One of the plan’s findings shaped the whole approach: employers screen far more on their application-form questions than on CVs. So the plan says three CVs, not one per employer, with the time going into good form answers instead. A second AI, from another company (OpenAI’s Codex), reviewed the CV draft as well.

Almost every screen in the app comes from something in that research:

In the researchIn the app
A, B and C lists, plus a few long shotsA tier on every opportunity, shown as a bold coloured label, and a To decide tab that puts the best tiers first
Three CVs, one per kind of roleThe CV to use on every opportunity, filled in automatically when an application starts
The sheet’s “Status at 9 Oct 2026” columnA listing status (Open, Rolling, Opens later, Watch…) and the date it was last checked, with a Needs re-check list after 14 days
Deadlines written every which way: “rolling”, “expected Jan 2027”, “opens w/c 19 Oct”The employer’s wording kept word for word, and a closing date only when the text gives a full date
The sheet’s orange and yellow urgency coloursThe same colours, measured from today: orange within 14 days, yellow within 30
Each employer’s selection stagesA one-page prep sheet before every test and interview
The cover letter template and form answersAn answer bank with word limits and a copy button
03

A brief written for an AI

The research then became a 475-line brief for the app, plus a page of house rules for the AI: the data, every screen, how the import works, what must never happen, and the order to build it in. It’s precise in the places where software usually goes wrong:

Dates

Set closes_on only when the cell is a date, or the text starts with a full day-month-year date… Never guess a year. Always keep the original text.

Importing again

A re-import may update research fields… It must never touch [the graduate’s] own data: applications, status history, steps, documents, answers, contacts, event status, or dismissals.

Status history

Written automatically every time an application is created or its status changes, however the change is made. Use a model observer so no code path can skip it.

The house rules covered the rest: UK English, dates written “9 Oct 2026”, 24-hour times, small saved steps, never show a password or key in the chat, and anything on the server that only an account owner should do is left to Mark, with exact instructions. Then one prompt started the build:

  1. Mark · 21:09

    This folder is the whole project. Read CLAUDE.md and docs/BRIEF.md, then build Phases 0, 1 and 2 in a single pass without stopping for review, and finish with the handover in BRIEF §11. … Only stop if something fails that you can’t fix yourself, and tell me exactly what you need.

04

56 minutes to a live app

Claude Code, Anthropic’s coding tool, worked through the brief on Mark’s Mac, saving its work to GitHub as it went. About ten of the 56 minutes were spent waiting for things only Mark should do: creating access to the server and typing in the database password. When Claude came across a login with access to the whole server, it didn’t use it, and asked for one that could only reach this app.

  1. 21:09The prompt is pasted in.
  2. 21:11No access to the server yet, so Claude sends Mark a notification and carries on building on the Mac.
  3. 21:20Phase 0 saved: the app’s skeleton and its sign-in page, hidden from search engines.
  4. 21:22The data model, every screen, the dashboard, the agenda, a drag-and-drop board, documents and the importer.
  5. 21:39The real research imported on the Mac: 260 opportunities, 12 events and 415 sources. Importing it a second time changes nothing, as the brief asked.
  6. 21:43It clicks through every screen in a browser, at laptop and phone sizes: start an application, move it on, drag a card.
  7. 21:49Phase 2: reminder emails, a calendar feed and a stats page. 119 tests pass.
  8. 21:56The server runs an older version of PHP, so Claude moves the app back one version of its framework. All 119 tests still pass.
  9. 22:02Code on the server, database set up, and the research imported there too.
  10. 22:03The tests run again on the server’s own PHP, in a throwaway copy, so the live install isn’t touched.
  11. 22:05Live, with a handover note: what’s been built, the import summary, and Mark’s own steps, with the exact commands.
05

Then, a conversation

With the app live, Mark used it, and asked for changes in plain English as they came up. Claude built each one, tested it and put it live, usually within minutes. Some of that evening and the next morning:

  1. Mark · 22:29

    So how can I update the app with new entries if I find them, preferably through your research if possible? Can’t you just automatically check the live Google Sheet?

  2. Claude · 22:30

    Two quick questions, each with a recommendation: let a small script inside the sheet send it to the app (it keeps the sheet private), and keep the sheet as the place new research goes. Mark picked both. The sync was live at 22:35.

  3. Mark · 22:50

    Let’s make the tiers’ colour coding better: bright green for A, blue for B, amber for C. Make these a bigger font and bold. … This applies wherever the tier is shown in the app.

  4. Claude · 22:52

    Done, everywhere a tier appears.

  5. Mark · 23:33

    If I’m not applying for a job and it’s marked as such, put an obvious indicator. … Also put it in the listing itself: big red message.

  6. Claude · 23:41

    Red rows and a red NOT APPLYING label in the lists, and a big red banner on the listing with the date and the reason.

  7. Mark · 23:43

    Let’s just add an obvious way on the lists to show or hide “not applying” opportunities. Remember the choice.

  8. Claude · 23:44

    Done: a Show/Hide button that remembers each person’s choice, and turns red with a count of what’s hidden.

  9. Mark · 08:27, next morning

    After you’ve finished, can you carry out a full audit of the app, checking for bugs, and look for improvements?

  10. Claude · 08:35

    Three independent reviewers, working in parallel on security, correctness, and usability with server risks. Every finding checked against the code before anything is fixed. Then seven ideas for improvements.

  11. Mark · 08:38

    Can we also add a help section? … These instructions will be kept up to date by you as we make changes to the app.

  12. Mark · 08:39

    Also add little i icons next to anything ambiguous. … Language needs to be really clear.

  13. Claude · 08:48

    A Help section: a 2,500-word manual, the job-hunt strategy, a glossary of 34 terms and about 70 small ⓘ explanations around the app, with a standing rule that every change updates them.

  14. Mark · 09:20

    Do all ideas.

  15. Claude · 09:41

    All seven built, tested and live.

Mark’s own messages, with typing slips tidied. Claude’s replies are summarised.

06

Every application, in one place

Opportunities are the research: every scheme with its tier, the kind of work, where it is, the closing date in the sheet’s urgency colours (or the employer’s own words when there’s no date) and the CV to send. Tabs split them into Active, To decide, Applying, Not applying and Closed, and the filters above the list apply as you choose them.

Start an application and it opens at Preparing with the right CV filled in. From there it moves on through Applied, Online tests, Interview, Assessment centre and Offer, and every move is written into its history automatically, however it’s made, so the timeline on each application is always complete.

The Opportunities list on its To decide tab: filters for region, tier, fit, listing status, country, closing date and application, then rows with a green A or amber C or red Stretch tier, the kind of work, employer, programme, location, an Open or Rolling label, the closing date in orange or yellow with “in 12 days”, and the CV to use

To decide

Open schemes with no application yet, best tiers first, then the soonest closing.

  • Opportunities
An application page for a made-up employer, Cobalt Vision: buttons to move it on, mark the next action done, open the prep sheet, the apply link and the portal; its status, Interview; the date applied; the CV used; the next action and when it’s due; and a timeline on the right from Started to Interview

An application

Its timeline merges every status change and completed step, newest first.

  • Applications
The Move to… form over an application: the new status, the date it happened, a note, and an Add it as a step switch turned on, with fields for the type of step, when, the time and where or a link

Move to…

Moving to a test or an interview offers to add it as a step, with the date, time and link, so it lands in the agenda and the reminders.

  • Applications
The Board: six columns, Preparing, Applied, Online tests, Interview, Assessment centre and Offer, each holding cards for made-up employers with the programme, the closing date, the next action and Move to… and Done buttons

The board

Drag a card to the next column and the Move to… form opens with the new status filled in, so the history is still recorded. Built with plain Alpine.js, no plugin, and it fits six columns on a laptop.

  • Board
  • Drag and drop
A listing for a made-up bank marked not applying: a red NOT APPLYING label beside the name and a big red banner, “Marked as not applying on 10 Oct 2026. Reason: Prefer engineering roles to banking.”, above the listing’s details
The same listing on a phone, the red banner under the buttons

Big red message

A decision not to apply can’t be missed: red in every list, a banner on the listing, and left out of the reminder emails.

  • Opportunities
07

Ready for the next step

The agenda shows the next eight weeks, week by week: closing dates, tests and interviews, careers fairs and next actions, each with its status, tier and listing labels, and a colour key at the top. It’s built on one shared list of what’s coming up, which also feeds the dashboard, the reminder emails and the calendar feed, so they always agree.

Before a test or interview, the prep sheet puts everything on one printable page: what’s coming up and the notes for it, the employer’s selection process, eligibility, research notes and links, the answers used on the form (with copy buttons), contacts, documents sent and the steps done so far.

The Agenda: buttons to hide not-applying items and show closed listings or the past two weeks, a colour key of every status, then This week and Next week, each a table of dates with status, tier and listing labels and what’s happening
The agenda on a phone, the labels sitting in front of each item

The agenda

Eight weeks ahead, with buttons for the past two weeks and for listings that have closed.

  • Agenda
  • Phone
A prep sheet for an interview with a made-up employer: At a glance with the status, closing date, date applied, location, salary, start date, CV sent and portal; Coming up, a technical interview with its link and preparation notes; and About the role, with the selection process, eligibility and research notes

The prep sheet

One page to read before an interview, linked from the agenda, the calendar and the emails.

  • Interviews
Editing an answer in the answer bank: the question “Tell us about a technical project you are proud of.”, a long answer, and under it in red “132 words · 770 characters · 12 over the 120-word limit”, with themes and the word limit below

The answer bank

Good form answers, written once and reused. The word count turns red as soon as an answer runs over the limit.

  • Application forms
Documents: buttons for the CV folder and the sent CVs folder in Google Drive, then a library with a slot for each of the three master CVs and the cover letter template, each with an Open in Google Docs button, and an uploaded PDF on the first

Documents

CVs are edited in Google Docs. Each application keeps a link to its own copy and the PDF that was actually sent.

  • CVs
  • Google Docs
The Stats page: applications per status, how many applications reached each stage from Applied to Offer, overall, by kind of work and by tier, and a bar chart of applications sent each week

Stats

How far applications get, worked out from the history, so a rejection after an interview still counts as reaching interview.

  • Stats
08

Reminders that come to you

Nobody should have to open an app to find out they’ve missed something. At 7:30 each morning, if anything is due in the next three days or is overdue, an email lists it. On Mondays a longer one covers the next two weeks, where every application stands, anything that’s gone quiet and how many listings need checking again.

There’s also a private calendar feed for Google, Apple or Outlook calendars, with six months of closing dates, tests, interviews, events and next actions. Each person can switch the emails off, and get a new feed address if the old one is ever shared by mistake.

The Monday reminder email on a phone: “Hi Alex, Here’s what’s coming up over the next two weeks.”, an Overdue heading in red with one item, then Tomorrow, Monday and Tuesday, each with linked items such as a coding test and a careers fair

The Monday email

Plain, friendly UK English, every item linked to its page.

  • Email
The Profile page: name, email address and a new password field, an Email me reminders switch with its explanation, and a Calendar feed address with Regenerate and copy buttons

Reminders and the calendar feed

Switch the emails on or off, and copy the feed address into any calendar.

  • Calendar
  • Email
09

The Google Sheet, synced every morning

The sheet didn’t go away. It’s still where research goes, by Claude or by Mark, and the graduate never needs to open it. The app is where the job hunt happens. Every morning at about six, a small script inside the sheet sends a copy of the whole thing to the app. The sheet also has a menu: Send to app now, and Check what would change, which changes nothing and says what would happen.

Google Sheet + 6 new rows from research Tracker · Abroad · PhD · Events 06:00 secret key · HTTPS checked, planned, applied in one go Job Applications Your applications: untouched 6 new · 0 updated · 260 unchanged

How the sync works. The figures are from the first sync after a research run.

  • Columns are found by their headings, not their position, so a column moved in the sheet doesn’t break anything. If one is missing, the pop-up in the sheet says which.
  • It plans before it changes anything. The whole sheet is read and every change worked out first, then applied in one go. If anything is wrong, nothing changes.
  • New rows become new opportunities, changed rows update them, with one exception: if someone changed that detail in the app since the last sync, the app’s version wins, and the report says the two disagree.
  • The graduate’s own work is never touched: applications, interviews, notes, documents, answers, contacts, event bookings and decisions not to apply.
  • Nothing is ever deleted. A row removed from the sheet stays in the app and is listed in the report.
  • Dates are never guessed. “Expected Jan 2027” stays as words. Only a full date becomes a closing date.

How the app decides between the sheet and its own edits, from the demo copy, with made-up employers. It keeps a copy of what each sync wrote, and compares three versions of every field:

FieldWritten by the last syncIn the app nowIn the sheet nowWhat happens
Salary, Fusion Futures Lab£35,000£35,000£36,500Nobody changed it in the app, so the sheet’s new figure goes in
Notes, Meridian SoftwareSmall intake; apply early.Spoke to them at the careers fair: apply before December.Rolling: most places go before Christmas.Changed in the app, so the app’s note stays, and the report shows both

Job Applications · Check what would change

Dry run, nothing saved.
Opportunities: 1 new, 4 updated, 26 unchanged.
Events: 0 new, 0 updated. Sources: 1 new.
1 changed in the app, so kept (see the report).
Deadline calendar conflicts: 1. Skipped rows: 0.

The summary the sheet shows after a dry run, from the demo copy. A full report is saved on the server each time.

The sheet stays private: it’s never shared by link. The script runs as Mark and proves itself to the app with a secret key, which the app checks before it reads anything. Only one sync runs at a time, the file must really be a spreadsheet and under 10 MB, and a stray value far down the sheet can’t make it run out of memory.

10

Checked by AI, several ways

Nobody read every line of this code, so it had to be checked in other ways. Claude wrote tests alongside every feature and ran them before every save: 22 when the first features were saved, 234 by the end. It looked at every screen in a real browser after each change, about 60 rounds of checks at laptop and phone sizes. And the morning after launch, Mark asked for a full audit.

Claude set three separate reviewers to work at the same time, each a fresh AI session with the project’s rules and one job. They could read every file and run the tests, but not change anything, and every finding needed the file and line as evidence. The main session then checked each finding against the code before fixing it.

  1. Security and data safety

    Every route into the app, every piece of hand-built HTML, uploads, links, password reset and the importer as a way in. 3 medium and 9 low findings.

  2. Correctness

    The import’s matching and merging, the history, reminders, time zones in the calendar, the stats and slow database queries. 2 high, 4 medium and 10 lower findings, each bug reproduced in a scratch database first.

  3. Usability and the server

    Labels, empty screens, phone layouts, gaps against the brief, and every PHP function a typical server switches off. 14 usability points, 15 server risks and 8 ideas.

A few of the things they found, all fixed with tests the same morning:

  • A link typed into the sheet could have run code in the app. Now only web addresses become links.
  • A forged request could have pointed password-reset emails at another website.
  • The forgot-password page gave away which email addresses had accounts.
  • Changing a row’s link in the sheet would have created a second copy of an application.
  • Overdue next actions dropped off the dashboard and the emails.
  • The board asked the database a separate question for every card: 242 queries for 60 cards.
  • Spreadsheet cells that look like formulas are made safe in CSV exports.
  • The deploy script could have left the site down if the connection dropped half way.

php artisan test

............................................................................
............................................................................
............................................................................
......


Tests: 234 passed (1019 assertions)
Duration: 10.74s

The app’s tests, run on the demo copy on 10 October 2026. Each one starts from a fresh, empty database.

11

The stack, on our own server

The app runs on Clifton Media’s own server, behind a sign-in, and tells search engines to stay out. There’s no public sign-up: Mark creates every account directly on the server, so passwords never pass through the AI. Files live on a private disk and download only for someone signed in, and keys are kept out of the code, the chat and the logs.

Each update goes live with one command: the site shows a short “back in a moment” page for about 20 seconds while the new code goes in, and a safety net brings it back up if any step fails.

PartWhat it isWhat it does here
Laravel 12A PHP framework for web appsThe app itself: data, rules, emails, the scheduler and the sync
Filament 5Ready-made admin screens for LaravelLists, filters, forms, tabs and the dashboard, with no third-party plugins
Livewire and Alpine.jsSmall tools for screens that update without reloadingLive word counts, the drag-and-drop board, instant filters
PhpSpreadsheetReads Excel files in PHPReads the sheet, including real date cells
MariaDBThe databaseEverything the app stores. The tests use SQLite instead
Google Apps ScriptScripts that run inside Google SheetsThe 6am sync and the sheet’s menu, in 92 lines
SendGridEmail deliveryReminder emails and password resets
PestA testing tool for PHPThe 234 automated tests
GitHubWhere the code is keptEvery change saved with a note of what and why: 36 in all
CloudwaysManaged cloud hostingOur server, with HTTPS and a cron job that runs the scheduler every minute
12

Which Claude, and how hard it thought

Anthropic makes Claude in several sizes, and you choose one for each job. The bigger models are better at long, complicated work and cost more to run. You also choose an effort level, from low to max: how much the model thinks before it acts. More effort is slower and uses more tokens, but makes fewer mistakes over a long build.

Every step of this build ran on Claude Opus 5.5 at “xhigh” effort, the second-highest setting, and so did the three reviewers.

ModelWhat it’s forOn this project
Claude Fable 5.1Anthropic’s most capable model, and the most expensiveNot used for the build
Claude Opus 5.5Long, complicated work such as building softwareThe whole build, the fixes and the three reviewers
Claude Sonnet 5.5Quicker and cheaper, for lighter jobsTwo research sessions looking for new graduate schemes
Claude Haiku 5.5The smallest and quickestSummarising web pages while researching

AI models read and write in tokens, a few letters each. The build wrote about 870,000 tokens of code, reasoning and reports, and read about 375 million: mostly the conversation so far, re-read at every step from a cache, which costs a fraction of fresh reading. Claude Code’s own counter put the first two hours and twenty minutes at US$46.54 at Anthropic’s pay-as-you-go prices.

13

My take on AI

I didn’t write a single line of this code, and I couldn’t tell you in depth how most of it works. In that sense it’s a black box. What gives me confidence is how it was checked: Claude tested its own work as it went, and separate AI sessions audited it independently, each looking for different problems, with every finding checked before it was fixed.

It’s like a bunch of boffins coming together and building it to best practice.

Mark Pollak, founder, Clifton Media

Could we release it publicly? I’m not sure. It was built for one household: there are no user roles, everyone sees everything, and it expects one particular Google Sheet. A public version would need support, and there’s the question of liability if it were ever hacked. Then again, that can happen with software written by people too.

Working at this level burns through quite a few tokens. But it’s still amazing: a proper bespoke tool, researched, built, checked and live in an evening. That just wasn’t possible before.

Results

The app went live on the evening of 9 October 2026. The first research run after launch added six new verified schemes. A longer search the next morning checked about 250 employers’ careers pages and found nothing new, a good sign the original research was complete.

One idea is waiting for later: paste in a job advert, and have Claude fill in the new opportunity for checking.

By the end of the second morningHow many
Opportunities in the app266
Saved changes to the code36
Automated tests234
Lines of application code9,226
Words of help inside the appAbout 4,000

From the build notes, 9 and 10 October 2026.

Built with

  • Claude Code
  • Claude Opus 5.5
  • Laravel 12
  • Filament 5
  • Livewire
  • Alpine.js
  • PhpSpreadsheet
  • MariaDB
  • Google Apps Script
  • Google Sheets
  • SendGrid
  • Pest
  • GitHub
  • Cloudways

Built for one family’s job hunt, not for sale. If you have a spreadsheet that’s trying to be an app, we’d like to hear about it.

NextTopoLineA climbing topo app that works with no signal: draw your routes, filter by grade, use it at the crag. A work in progress.

Have a project in mind? We would like to hear about it.

  • Book a callThirty minutes on a video call, at a time that suits youBook a call
  • Phone07515 730686
  • Email
  • BasedMatlock, Derbyshire