T.A.L.Q.T.
  • overview
  • features
  • specs
  • the markdown
  • manual
  • guides
  • blog
  • deutsch
download

blog 05 · plain text

why a markdown time log beats an export.

One file per day, forty lines, readable by anything. The argument for keeping your working time as text, and the two places where a database still wins.

published 8 November 2026 · felix barth

This is what a working day looks like once it is over. A file called talqt-2026-11-08.md with four parts:

  1. Overview: target 8:00, worked 8:12, break 0:45, presence 8:57.
  2. Timeline: a table with one row per segment, columns start, end, duration, mode (work or break).
  3. Task log: what was said into the recorder, each line with its time stamp.
  4. Time summary: the day's totals once more, as text.

About forty lines, plain text. It opens in Obsidian, VS Code, Bear, TextEdit, a terminal, a browser. It will open in whatever exists in 2046. This post is about why that format is a better home for a working-time record than a database you can export from, and about where it falls short.

01 · ownership is a file, not a feature

Most time trackers store your entries in their database, on their server, and offer an export. The export is a feature. Features change, move behind a higher plan, or disappear with the company. Your record is also only as current as the last time you remembered to export it.

A file is different in kind. If the record is written to your own disk or your own iCloud Drive on the day it happens, there is nothing to export. The record already is the file. Cancel the app, lose the login, let the company fold: the folder is still there, and so are the 240 days in it.

This is why talqt writes one markdown file per day at checkout instead of keeping a database and adding an export button. The app keeps no copy anywhere else. No account, no server. The file in your iCloud Drive is the only record, which is both the feature and the responsibility. The privacy page says the same thing in fewer words.

02 · search that already works

Plain text is the one format every search tool understands. Spotlight on the Mac indexes the folder without being asked. Obsidian searches 500 daily files in the time it takes to type the word. On the command line, grep -l Müller talqt-2026-*.md lists every day in 2026 that mentions that customer, and grep -c counts the mentions. No query language, no report builder, no dashboard to wait for.

The question of which days in September went to the Müller renovation is a database query in a SaaS tool and a text search in a folder. Both work. Only one of them still works when the app is gone.

03 · a format that language models read natively

This argument did not exist five years ago. If you want to ask an AI assistant how many hours went to the Müller project in October, and on which days, you have to give it the records. A database export has to be converted first. A folder of markdown files can be handed over as it is. Markdown is what most language models were trained on in large quantities, and a table with start, end, duration and mode is something they read reliably.

That also means you can do this locally, with a model on your own machine, without uploading time records to anyone. The files are small. Two hundred working days of daily logs come to well under a megabyte. Whether you want to do that is a separate question. The format does not stand in the way.

04 · longevity

Text files from 1990 open today. Database files from 2010 often need the application that wrote them, in the version that wrote them, on an operating system that can still run it. For a record you have to keep for two years under § 16 Abs. 2 ArbZG, and might want to keep for a whole career, that difference matters.

Markdown adds a little structure on top of plain text: headings, a table, a list. If every markdown renderer vanished tomorrow, the file would still read fine as text, because the structure is visible in the characters themselves. A pipe is a column divider whether or not anything renders it.

A tangent, because it belongs here. Ships' logs were kept in bound books for four hundred years, one entry per watch, in ink. Nobody designed that format for retrieval. It survived because any officer on any ship could read any log, and because a page written at the time and never rewritten was trusted in court. The format outlasted the empires that used it.

05 · where the file approach falls short

A folder of daily files has no built-in aggregation. If you want hours per project per month, you sum them yourself: a script, a spreadsheet import, or an assistant as in section 03. A database gives you that in one click.

Renaming a project across 200 files is a find-and-replace, not a dropdown change. There are no relations: a task does not know it belongs to a customer unless you said so when you recorded it. And there is no multi-user view at all. Each person has their own folder, and merging them is a manual step. How small teams live with that.

For one person who wants a durable, searchable record of each day, these are acceptable costs. For a company that invoices 40 projects to 15 clients from 8 people's time, they are not, and a database-backed tool is the right choice. The spreadsheet comparison covers the middle ground.

06 · how the file gets written

In talqt the day runs like this. You set the start time on a clock face or take the current time, press START DAY. During the day you hold RECORD, say the task, release. The timer runs for that task until the next entry or a break. In the evening you check out, and the app writes talqt-YYYY-MM-DD.md into a folder you chose in your iCloud Drive, with the four parts from the top of this post. Open it in Obsidian and the table renders. Open it in TextEdit and you see the pipes. Either way the file is yours, and it was yours from the moment it was written.

07 · who talqt is for, and who it is not for

talqt is for one person who wants each working day recorded by voice and kept as a plain-text file in their own iCloud Drive. iPhone only, iOS 17 or later, one-time purchase, no account, no subscription, no server. It is a single-user app with no team view, no invoicing, no API and no Android or web version. If you need aggregation across people and projects at the press of a button, a database-backed tool is the better fit. If you want a record that will still open in twenty years, a folder of text files is hard to beat.

talqt on the App Store

faq

What is in a talqt daily file?

An overview (target, worked, break, presence), a timeline table with start, end, duration and mode per segment, a task log with time stamps, and a time summary. One markdown file per day, named talqt-YYYY-MM-DD.md.

Where are the files stored?

In a folder you choose in your own iCloud Drive. talqt keeps no copy on any server and has no account. The files are the only record.

Can I open the files without talqt?

Yes. They are plain text with markdown formatting. Obsidian, VS Code, Bear, TextEdit, Notepad or any text editor opens them. Markdown apps render the table; plain editors show the pipes.

How do I get monthly totals from daily files?

Sum them yourself: import into a spreadsheet, run a short script, or hand the folder to an AI assistant and ask. talqt has no built-in cross-day aggregation.

related reading

  • talqt vs. the Excel timesheetThe spreadsheet is free and adds up your hours. talqt records them live by voice. Where each one wins, and what you give up when you switch.
  • time tracking tips for small teamsTime tracking for teams of 2 to 10: one method for everyone, a daily close, plain text files, a weekly review. Honest about where a single-user app fits.
  • talqt vs. subscription time trackersTeam trackers bill per user per month and keep your data on their servers. talqt is a one-time-purchase iPhone app for one person. Where each side wins.
  • project time tracking, one switch at a timeHow to record project time so the totals mean something: granularity, timer versus reconstruction, and how talqt logs each switch with your voice.

→ in this section: blog

talqt

talk assisted quick tracking.
voice time tracker for iphone.
made in germany.

links

  • overview
  • use cases
  • compare
  • blog
  • manual
  • privacy
  • contact
  • app store

edition

iphone · v1.9
build · 2026.10
© talqt