Personal Productivity Tools in the Age of AI

For about a decade I’ve kept a quiet mental list of small productivity tools I wished existed: a Gantt chart tool that didn’t force me to register an account before drawing a single bar, a checklist app I could share with someone without making them register either, a private day log I actually owned and could trust to still be there in five years. None of these ideas were complicated, none of them needed a venture round to exist, and yet they stubbornly refused to materialise in the shape I wanted, because the off-the-shelf options were always either too bloated, too expensive, or a little too eager to swallow my data on the way in.

I never built any of them, and the reason is not particularly mysterious. I’m a biologist by training and an IT consultant by trade, which means I can read code reasonably well and write it badly under pressure, but the gap between “I have an idea” and “I have a working web app with a backend, a database, a deployment pipeline, and a frontend that doesn’t look like it was last touched in 1998” was always wider than the time I had available for hobby projects. Then AI happened, and that gap quietly closed.

The old math

The reason most knowledge workers don’t build their own tools is not that they lack ideas, it’s that the ideas die quietly in the cost-benefit analysis. A custom tool that saves you ten minutes a day sounds great in theory, but a custom tool that costs you three weekends of fighting with frameworks, learning a database you didn’t want to learn, and deploying something you don’t fully trust is much harder to justify in practice. So you keep paying the bloated SaaS subscription, you keep filing the irritation under “not worth it”, and you tell yourself that good enough is, in fact, good enough.

I lived inside that calculation for years. I’d open a Gantt tool to sketch a project plan, get hit with a sign-up wall on the third click, close the tab, and reopen Excel as if nothing had happened. I’d want to share a recurring checklist with a colleague and end up emailing them a Word document that we both knew neither of us would ever update. None of these moments were dramatic individually, but they added up to a kind of low-grade resignation about software, the quiet acceptance that the tools shaping my daily work would never quite fit the work itself, because fitting them was somebody else’s job and that somebody else had different priorities than I did.

The new math

What AI changed is not that I suddenly know how to write production code, but rather that I no longer need to in order to get something useful out of the door. With Claude Code and similar tools I can describe what I want in plain language, watch a working version of it appear within minutes, run it, find what’s broken or missing, describe the fix in another sentence or two, and iterate. The whole loop from idea to running tool comfortably fits inside an evening or two, which is the kind of timescale that makes “let me just build it” a reasonable response rather than a fantasy.

That probably sounds like a marketing line, so let me be more specific about what it actually feels like in practice. The first version of any of my tools is invariably rough around the edges, because the AI tends to get the structural decisions right and the small interaction details wrong. I use it for a day, write down five things that annoy me, describe them, get a refined version back, and use that for another day. After about a week the tool has been reshaped to fit my workflow with surprising precision, mostly because I am simultaneously the only user, the only product manager, and the only person whose feedback matters. After a month I’m using it daily and have stopped thinking about it the way I think about other software: it has receded into the background and become more like a hammer than like an app.

The wild part is what all of this costs. The combined hosting bill for the three tools I’m about to introduce is a few euros a month, I own the code and the data outright, and I can change anything I want whenever the workflow underneath shifts.

gantt.guru

I needed a Gantt chart tool I could open, sketch a project plan in, and share with a client without either of us creating an account first. Every existing option I tried either demanded registration before showing me anything useful, or buried the chart inside a full project management suite that came with twenty other features I had no intention of using.

gantt.guru gives each project two URLs, one for editing and one for read-only viewing, and you share whichever fits the situation. Tasks support subtasks and the usual dependency types, the timeline is zoomable, and finished plans export cleanly to PNG, SVG, or vector PDF for slide decks. Each project also has its own MCP server, which means a GenAI tool of your choice can read and modify the plan directly without me having to glue anything together by hand. I end up using it most weeks for client work.

Checklister

I run a handful of recurring processes, things like release checks, onboarding flows, and weekly reviews, and I wanted to be able to define each checklist once as a template, spin up clean instances from it whenever a new run started, and share each instance with whoever needed it without forcing any of us to log in somewhere.

Checklister is built around exactly that distinction between templates and instances. A template defines the items, each run produces a fresh instance derived from it, and access is controlled entirely through tokens embedded in the URLs: an admin token for the template itself, plus three separate tokens per instance for checking items off, editing them, and read-only viewing. You hand out whichever token matches the role the recipient should have, and there are no accounts involved on either side.

Personal Timeline

I wanted a private place to log what I had actually done each day, not a journal in the introspective sense and not a notes app in the everything-bucket sense, but a chronological microblog I controlled, with hashtags so I could find things again later and immutability on past entries so the log stayed honest with me over time.

Personal Timeline is a single Go binary that bundles the frontend and stores everything in SQLite. Hashtags are extracted and indexed automatically, editing is restricted to the same calendar day an entry was created, and there’s an API key mechanism for posting from scripts as well as an outbound webhook that fires whenever a new entry lands. The detail I’m secretly proudest of is that each past day is cryptographically sealed and the resulting hash is committed to OpenTimeStamps, which makes the log tamper-proof against the Bitcoin blockchain. That is wildly excessive for a personal diary, and I love it precisely because of that.

What this actually shifts

The interesting thing about all three of these tools is not the tools themselves but the fact that none of them would have existed two years ago, and not because they were technically hard to build, but because the cost of building a small, focused, opinionated tool was simply higher than the cost of putting up with somebody else’s product. That ratio has now flipped, and the consequences of the flip are still working their way through how knowledge workers relate to their own software.

For someone like me this is genuinely new territory. I’m not a software engineer and I have no particular ambition to become one, but I do carry around a decade of accumulated frustration with tools that were almost-but-not-quite right, and I now have a way to translate that frustration into something I can actually use. The tools are imperfect, the code under the hood is not pretty, and I would not let any of them anywhere near a regulated environment or sensitive data I didn’t fully control. For my own daily work, however, they fit exactly the way I always wanted productivity software to fit, and that experience has been genuinely fantastic in a way I didn’t quite expect.

If you’ve been carrying around your own private list of “I wish a tool existed that just did X”, this is a good moment to look at it again with fresh eyes. The reason that tool doesn’t yet exist is probably not that nobody else wants it, but that nobody else wanted it badly enough to push through the old cost-benefit calculation, and that calculation no longer holds. What used to be a barrier is now, at most, an evening.