Free download for macOS

Pip thinks with you,
not for you.

Most AI tools produce output. Pip takes the operational noise off your plate — Slack, email, meetings — and when something needs working through, he thinks alongside you.

Download for Mac
Pip — Home

Morning.

Your quarterly review is on Thursday. Three of the five sections still need shaping — I've blocked an hour tomorrow so you can sit with them.

Sam tagged you in #planning yesterday about the new pricing tier. He's asked for your take before Wednesday's call. I've drafted a reply — want a look?

From Tuesday's product sync, two open actions never made it to the tracker. Both look medium effort.

Chuck something at Pip…

Needs actioning

Reply to Sam re: pricing tier — draft ready
Shape "metrics" section for Thursday's review
Confirm slot for Friday's customer call
our belief

The value isn't in the output. It's in the thinking that gets you there.

— how Pip is built

thinking partner

When you want to think something through.

When you're working through a decision, Pip shifts. He stops being the doer and starts being the person who asks the question you haven't got to yet.

He surfaces context you might have missed. Challenges reasoning with evidence, not opinion. Pushes back when something doesn't add up. And knows when to stop — because the conclusion is always yours.

in your notes

He reads as you write.

When you're drafting in your notes, Pip reads along. He surfaces questions in the margin — to sharpen what you're building, not to dictate it.

Pip · Notes / Q3 strategy

Q3 strategy — early thinking

draft last edited just now

We should double down on enterprise. The mid-market segment has been underperforming for two cycles. Time to refocus.

Three deals look real right now:

  • Acme Co — enterprise, in procurement
  • Tessa Inc — mid-market, stalled
  • Northwind — enterprise, qualifying

If we get Acme and Northwind across the line, that covers H2 targets. Mid-market can wait.

pip is reading
Quick one — "underperforming" against what baseline? The targets you set in January look high in hindsight.
Worth flagging: Sam's team has been chasing Tessa for three cycles. He'll want a heads-up before this lands.
You committed to mid-market progress at last month's all-hands. Worth pulling that recording before the cycle review?
gets things done

Everything else, handled.

The thinking only works if you have time for it. Pip handles the operational load that eats that time — without being asked.

Catches what's falling through

Monitors Slack, flags obligations going stale, chases loose ends before they disappear.

Preps you for meetings

Spots meetings that need prep and surfaces what you need before you walk in.

Works from real conversations

Drafts from your actual meetings and threads — not generic templates.

Connects your tools

Pull from Granola, draft, push to Notion, flag in Slack. Each tool stays in its lane.

integrations

Your tools, connected.

Pip doesn't replace any of these. He's the connective tissue between them.

Slack Notion Gmail 31 Google Calendar G Granola Linear
on your machine

Your data stays here.

Pip lives on your computer. Your notes, your context, your history — saved on your hardware, not ours. The trust holds because it's specific.

01 / local-first

Stays on your machine.

Everything Pip knows about you — your notes, your history, your context — is saved on your computer. Nothing leaves unless you explicitly connect a service.

02 / your choice of AI

Pick a model. Any model.

Plug in Anthropic, OpenAI, or Google. Or keep everything completely private with a local model that never leaves your computer.

claude-sonnet-4-5
gpt-5
gemini-2.5
llama 3.2 (local)
deepseek (local)
+ custom
questions

Things people ask.

Yes, while we're in beta. We're figuring out the model. When we land on one, you'll hear about it before anything changes — and there'll always be a free, local-only tier.
macOS only for now. Windows and Linux are on the list but we want Mac to be excellent before splitting our attention.
An LLM — large language model — is the bit that does the actual thinking. Claude, GPT, Gemini are the well-known ones. They read what you say, weigh it against everything they've been trained on, and write a response. Pip is the layer that connects an LLM to your tools, remembers context, and decides when to act.

It matters because different models think differently. Some are better at writing, some at reasoning, some are faster, some are cheaper. And some run entirely on your machine — slower, but nothing ever leaves. Pip lets you pick the one that fits the work, and switch any time.
On your machine. Your notes, your history, your context — all stored locally. The only things that leave are tool calls you've connected (Slack, Gmail, etc.) and prompts to whichever LLM you choose to route through.
Yes. Llama and DeepSeek both run on your machine — free, private, slower than frontier models, but nothing ever leaves your computer.
Short, warm, a bit of a name and a bit of a noise. He's a character, not a product feature.
Just me — Hayden, in Brighton, UK. I was tired of AI tools that confidently wrote things I didn't ask for, so I'm building the one I actually wanted.

Try Pip.

The thinking is yours. The admin isn't.

Download for Mac