The Tools That Survived My Workflow in 2025

Most tools I try do not make it to the end of the quarter. A few make it to the end of the year and become part of my permanent setup. Here is the short list of tools that earned their keep across a full year of real work.

Development

Claude Code: CLI agent, daily driver
Cursor: IDE agent for single-file tight loops
VS Code: fallback editor
iTerm2: terminal
ruff: Python linter and formatter
tsc: TypeScript compiler, always strict mode

Infrastructure

AWS Lambda: all serverless compute
DynamoDB: all database work
SES: all transactional email
Vercel: all frontend hosting
GitHub: git hosting and Actions

Business

Linear: task tracking
Notion: client docs and internal wikis
Harvest: time tracking for hourly work
Stripe: invoicing and payments

Writing

Markdown files in git: source of truth
Custom artifact scripts: batch publishing
Browser devtools: preview before publishing

The Pattern in the Survivors

Every tool on this list shares three traits:

Does one thing well: no suite software, no kitchen sinks
Predictable pricing: monthly cost known in advance
Reliable API or CLI: I can automate it or replace it

Tools that failed one of these dropped off my workflow somewhere during the year.

What I Did Not Add

No AI note-taking apps stuck
No project management SaaS beyond Linear stuck
No extra browser extensions stuck
No AI chat apps beyond Claude stuck

The test for any new tool: do I open it the next day? If no, it is gone within a week.

The Subtraction Game

Productivity is not about adding tools. It is about the tools you stop using. Every tool in a workflow is a context switch. Fewer tools means fewer context switches means more output. The survivors above earned their place by being essential, not by being interesting.

The Budget Line

Total software cost for my solo operation: under $200 per month including AWS. That covers hosting, development, business ops, and writing. Running a technical consulting practice does not require a big software budget. It requires the right software budget.

The Replacement Test

Before adopting a new tool, I ask: what does this replace? If the answer is nothing, I probably do not need it. If the answer is clear, the new tool has to be substantially better than the incumbent to justify the switching cost. That test kept my stack stable through a year of hype cycles.

The Upgrade Discipline

I upgrade my core tools on a predictable cadence: monthly for editors, quarterly for frameworks, annually for operating system. Out-of-band upgrades only happen for security patches. That rhythm prevents the all-at-once upgrade week that breaks everything simultaneously.

For pricing predictability, see the AWS pricing calculator to model your own infrastructure spend.