Easy, secure, local agents

Margaret's helpersget things done

Turn the work you repeat into small, private AI helpers — saved on your own computer, run on your terms.

Margaret's dashboard showing helpers, recent runs, and token spend

Watch · 60 seconds

See what a helper actually does.

A one-minute tour: what an “agent” really is, and how a single helper quietly takes a repeated task off your plate.

Title card reading “What’s an agent?”

Captions are burned in. Read the transcript.

AI is useful. Rebuilding the same prompt every time is not.

Most people use AI by pasting context into a chat window, tweaking the same prompt, and hoping the result is consistent. Agent builders promise more power, but often feel made for technical users.

Margaret gives you a simpler middle ground: small reusable AI helpers for the jobs you already repeat.

From messy task to reusable helper in minutes.

1
Create a helper

Tell Margaret what you want help with. She turns it into a reusable helper with inputs, instructions, and an output format.

2
Give it an input

Paste in text, fill a short form, or point to a local file — the same way every time.

3
Get your output

Margaret runs your helper and saves the result to your computer as a readable text file.

4
Improve over time

Ask Margaret what's vague, missing, or unclear. Apply suggested edits, then reuse the better version.

Create helpers for the work you repeat.

Helpers, not prompts

Write the instructions once in plain English. No re-explaining every time, no copy-pasting the same prompt into a chat window. Margaret remembers how you like things done.

Runs locally, saves locally

Everything lives in a folder on your Mac. Your helper library, your inputs, your run history — all readable files on your own machine. No cloud sync, no accounts, no subscription required for the app itself.

You stay in control

Nothing happens automatically. You decide when Margaret works. Before any run, you can see exactly what will be sent to the AI. You can edit, cancel, or improve any helper at any time.

A look inside

Built to feel calm, not clever.

A quiet, local desktop app where your helpers live. Dark by default, legible always.

Margaret's first-run welcome screen inviting you to describe a task
01

First run

Start from a blank page, or a single sentence.

Describe a task you spend time on and Margaret drafts a helper for you — or build one yourself. No setup ceremony, no empty dashboard staring back.

A Margaret helper detail view showing model choice and input fields
02

Helper detail

Every helper, fully in view.

Pick the model, see the cost, fill the inputs, and run. The instructions you wrote once sit right there — readable, editable, yours.

Margaret's tune-up panel suggesting edits to a helper
03

Tune up

Improve a helper without starting over.

Tell Margaret what to change and she revises the instructions in place. Your helpers get sharper the more you use them.

A Margaret workflow chaining several helpers together
04

Workflows

Chain helpers into a repeatable run.

Connect several helpers into a workflow for the work your week repeats — research into brief, notes into actions, transcript into summary.

Local-first

Built for people who care where their work goes.

Margaret stores your helpers, inputs, and run history on your computer. When a helper needs AI, Margaret sends the required content to the model provider you choose — using your own API key.

  • Your helper library lives on your machine
  • Your API key is stored securely in your system keychain
  • You can inspect what a helper will send before running it
  • You can open, back up, and edit your Markdown files yourself

Margaret is local-first, not magic. If you use a cloud AI provider, the text needed for that run is sent to that provider. We're upfront about this — it's in the app before every run.

Margaret's setup screen connecting an AI provider with your own API key

See it in action

Helpers become workflows.

One helper can summarise a call. Another can spot missing information. Another can draft a response. Used together, they handle the work your business repeats every week.

See all use cases →
01Sales enquiries

Qualify vague enquiries, find missing information, draft replies, and create lead notes.

02Support conversations

Summarise issues, classify urgency, flag risks, draft replies, and find FAQ gaps.

03Client delivery

Turn call notes into summaries, action plans, follow-ups, and scoped next steps.

04Workflow improvement

Map a repeated process, find friction, identify automation candidates, and build a safer plan.

Start with practical helper bundles.

Bundles are small packs of ready-made helpers and workflows for common work. Install one, adapt it, and make it yours.

Browse bundles
Free
Starter Bundle

Clean up notes, draft emails, summarize documents.

Coming soon
Consultant Bundle

Client intake, proposal outlines, recommendation memos, follow-ups.

Coming soon
UX Research Bundle

Research plans, interview guides, transcript synthesis.

Coming soon
Writing Bundle

Rough notes to outlines, blog drafts, LinkedIn posts.

Free to start. Yours to keep.

Free£0
  • Create and run local helpers
  • Starter examples included
  • Bring your own AI key
  • Local Markdown files
Download free
Margaret Plus
£49one-off

Early supporter price: £29 during beta

  • Workflow builder
  • Run history
  • Bundle management
  • Export and import helpers
  • Version history
  • Privacy receipts before runs
  • Better improve and apply tools

Common questions

Get launch notes and new bundles.

No spam. No enterprise nonsense. Just updates when something worth knowing ships.