Voice productivity stack: 9 tools we actually use to write, ship, and think

Loqua is a Mac-native voice typing tool that turns speech into app-aware text. A useful voice productivity stack needs more than transcription: it needs a place to think, a place to code, a place to track work, a place to capture meetings, and fast command surfaces for moving between them.

The stack principle

We rank Loqua first because it is the tool we wrote and the voice layer the rest of this workflow depends on. That is an honest framing, not a claim that every user should replace every tool with Loqua. Voice only becomes productive when the output lands in the right destination with the right shape.

The principle is simple: speak once, route cleanly. The voice layer should remove filler, preserve technical names, and format for the active app. The surrounding tools should keep the work moving without forcing the user back into manual copy-paste. A voice productivity stack succeeds when the time between a spoken idea and a saved piece of work is shorter than it would have been with the keyboard alone, and when the saved artifact is actually useful when someone (often future-you) opens it again.

The 9 tools

1. 1. Loqua

   Loqua is the capture layer for the voice productivity stack. We use it for Slack replies, Cursor prompts, code comments, Git commit messages, Linear issues, and long notes. The value is app-aware output: the same spoken phrase becomes a concise commit subject in a terminal and a warmer paragraph in Spark Mail.

   A representative week of usage: a few hundred dictated Slack and Linear messages, dozens of code comments and commit subjects, a couple of long brainstorming blocks dropped straight into Obsidian, and the occasional fast cleanup of a long-form draft. Loqua earns its place at the top of the stack because the friction between thought and saved text is consistently lower than typing.

2. 2. Claude Code

   Claude Code is where voice becomes implementation planning. We dictate the high-level change, then let the agent inspect files and propose edits. Voice works well here because prompts are longer than commands and benefit from nuance.

   A typical prompt we dictate runs four or five sentences: what we want done, which files to consider, what to avoid, and the success criterion. Typing a prompt that careful costs real seconds; speaking it costs almost none, and the agent loop benefits from the extra context. We routinely dictate the entire spec for a small refactor while looking at the diff.

3. 3. Cursor

   Cursor is our interactive code editor. Voice is strongest in Cursor when asking for targeted refactors, docstrings, code review notes, and test outlines. Loqua preserves identifiers from the visible file so prompts do not lose important names.

   The flow we use most often: select the function, hit the Cursor chat shortcut, then dictate "add a docstring with arguments returns and an example" or "refactor this to early-return on the null case." The screen context lets Loqua keep the function and parameter names intact, which is where typed prompts often break down.

4. 4. Obsidian

   Obsidian is where raw thinking becomes durable notes. Voice is especially good for daily logs, research summaries, and brainstorm trees. We dictate headings, bullets, and links, then clean structure with keyboard shortcuts later.

   A daily note in our vault often starts as a five-minute spoken stream: what we are about to work on, what is blocking, what we already promised someone today. Dictation makes the cost of capture low enough that the note actually gets written, which is the whole game with a knowledge graph.

5. 5. Granola

   Granola handles meeting notes. Loqua complements it by capturing pre-meeting context, follow-up tasks, and the five-minute post-meeting cleanup that turns a transcript into a plan.

   The handoff between the two tools matters. Granola gives us the rough meeting transcript; Loqua is what we use immediately afterward to dictate the three decisions, two follow-ups, and one owner-per-task summary that actually moves the work forward.

6. 6. Linear

   Linear is where spoken bugs become scoped work. We dictate title, context, reproduction steps, and acceptance criteria. The important move is separating "what happened" from "what done looks like."

   A useful voice habit here is dictating the acceptance criteria first, before describing the bug. It forces the spoken issue to end on a concrete definition of done rather than a vague complaint, which is the single biggest reason issues stall in review.

7. 7. Raycast

   Raycast keeps the stack fast. Voice produces the text; Raycast moves us to the right app, command, or snippet. The combination reduces the friction between spoken intent and the next concrete action.

   We pair Raycast snippets with Loqua for repeated patterns: a snippet sets up a Linear issue template, an Obsidian daily note, or a PR description skeleton; Loqua fills the variable parts with dictated text. The keyboard shortcut sets the shape, voice fills the content.

8. 8. Things 3

   Things 3 is our personal task capture. It is not a team tracker; it is the private place for "remember to check this after lunch." Voice makes capture low-friction enough that small tasks stop leaking.

   The reason voice matters more here than in a team tracker is that personal tasks are small, frequent, and easy to drop. If capture costs a typing interruption, the task gets skipped and reappears as a problem later. If capture costs a sentence into Things, the task survives.

9. 9. Spark Mail

   Spark Mail is where voice helps tone. We speak the rough answer, Loqua cleans it, and Spark sends the final email. It is faster than typing because customer replies often need warmth more than precision.

   An interesting pattern emerges with longer customer threads: the dictated text tends to be friendlier than the typed version. We have stopped trying to over-engineer that. If a warm sentence helps a customer feel heard, the time we spent dictating it instead of polishing a typed paragraph was well spent.

Comparison matrix

Cross-tool voice flows

What we tried and dropped

A working voice productivity stack is partly defined by what you remove. Three categories of tools we stopped using regularly: heavyweight project trackers that punished spoken input with mandatory fields, voice-first note apps that captured audio without producing reviewable text, and AI agents that needed a setup ritual before they could be useful. Each one looked good in isolation and broke the flow when added to a daily loop. The general rule we landed on: a tool earns a place in the voice productivity stack only if a 15-second dictation produces something a teammate could read on Monday morning without context.

How to choose your stack

Start with the destination where voice will save the most time. If you write many prompts, pair Loqua with Claude Code and Cursor. If meetings dominate your day, pair it with Granola, Linear, and Things. If you think in notes, start with Obsidian and Raycast. The best mac productivity stack is the one where voice output lands somewhere actionable.

The second principle is to keep the stack small until friction tells you to grow it. A stack of three tools that you use every hour is more productive than nine tools you remember to open twice a week. Loqua plus a notes app plus a tracker is a credible start; add the rest only when a recurring workflow makes the case for itself.