Inside the apiary
The apiary runs one quiet loop underneath the tools you already use: capture, distill, recall, compound. It notices what happens as you work, turns it into clean notes on Deeplake, and hands the right ones back the next time any assistant asks.
Capture
As you and your assistant work, a small local daemon records the important moments: what you asked, what the tool did, and what came back. It is cheap, instant, and out of the way. If it ever hiccups, your assistant keeps working.
Distill
Raw transcripts are long and noisy. The apiary distills them into a three-tier memory: a one-line key, a short summary, and the full original. Skim the headline, open the summary if it looks relevant, read the detail only when you must.
Recall
When you start a new session, the assistant gets a short briefing of what is already known, so it starts informed instead of blank. During work it can ask for more, matching both your words and the meaning behind them.
Compound
Every so often a self-tidying pass merges duplicates, drops the junk, and replaces stale facts with their current version, keeping full history. Memory gets sharper as it grows, not noisier.
From install to recall, in four steps
- 01
Install with one command
Paste one line in your terminal. It installs the local daemons and opens the hive dashboard in your browser. No Node, npm, or database setup first.
- 02
Connect your assistants
From the dashboard, click first time setup and approve the sign-in, then connect the coding assistants you already use. It takes a couple of clicks.
- 03
Work as normal
Keep coding. The apiary captures the useful moments on its own and primes each new session, so your assistant starts smart instead of blank.
- 04
Recall everywhere
A note written from one tool is recalled by another, on any machine you are signed in on, and across your team if you choose to share.
Stored on Deeplake, the database built for AI.
Every memory lives on Deeplake. It stores your notes as exact text and as vectors of meaning in one place, so recall can match words and match meaning at the same time. It keeps a full version history behind every fact, isolates teams and projects at the storage layer, and can run in your own cloud. Only the local daemon ever talks to it, and your secrets are stored apart and never shown to an agent.
Read about trust and securityCommon questions
How does the apiary capture my work without slowing me down?
A small local daemon records the important moments of each agent turn out of the way of your work. Capture is cheap and instant, and if it ever hiccups your assistant keeps working normally.
How does recall find the right memory?
It searches Deeplake by the words you used and by meaning at the same time, so it surfaces the right note even when you would never have guessed the exact term. It also primes each new session with a short briefing.
Does the memory get messy as it grows?
The opposite. An optional self-tidying pass merges duplicates, drops junk, and keeps the current version of a fact while preserving history. Recall gets sharper as it grows, not noisier.
What actually stores the memory?
Deeplake, the database built for AI. It holds your memory as exact text and as meaning at once, keeps full version history, and can run in your own cloud. Only the local daemon talks to it.
See the loop on your own machine.
Install the stack with one command and watch a note written today get recalled tomorrow, in a different tool.
Windows (PowerShell): irm https://get.theapiary.sh/install.ps1 | iex
Download