A personal AI chief of staff · runs on your own AI

It remembers,
so you can think.

Pinolo keeps your meetings, promises and projects — connected, traceable, and handed back to you in one calm brief a day.

How it works

Three acts, every day.

A chief of staff has one job: nothing falls through. Pinolo does it as a pipeline — capture the evidence, grow the memory, hand back only what deserves your attention.

01 · Capture

It listens where work happens.

Meetings, messages, forwarded email — pushed in by your tools or read by your own assistant. Everything lands as immutable evidence: the transcript as it arrived, never edited.

02 · Memory

It remembers with receipts.

Commitments, people, projects, goals — extracted, connected, and traceable. Ask any fact why it exists and the answer is a chain: this field, from that assertion, from that meeting, on that day.

03 · Calm

It comes back calm.

One brief in the morning that ranks brutally rather than lists exhaustively. Quiet hours. Hard caps. A “don’t remind me again” honored forever. Almost nothing earns an email.

What it actually does

The work it does unprompted.

A chief of staff doesn’t wait to be asked. These land on their own:

And when it doesn’t know your playbook — teach it.

A Pinolo skill is a page of plain language, not code. Whatever you teach it, it keeps — and shows you everything it saves.

“Save this as a skill.”

That claim procedure you just walked it through once? Next time it’s one sentence. Taught once, kept forever.

“Every night, read the team’s Slack and Linear. File what matters.”

Write that, schedule it, done — it runs while you sleep, with your assistant’s hands and Pinolo’s memory. No code, no middleman.

“Prep me for board calls the way I like.”

It remembers the recipe — per meeting, per person — and refines it every time you correct it.

Bring your own AI

The server that never thinks.

Pinolo runs zero language-model calls on its servers. The reasoning happens in the assistant you already use and already pay for — connected to Pinolo over MCP, an open protocol. Pinolo supplies the two things your assistant doesn’t have: a durable memory that can defend itself, and a disciplined background work loop.

No second subscription for intelligence. No copy of your credentials. See the wiring →

Structural, not promised

Built like it has something to protect.

Every guarantee below is enforced by structure — schemas, database roles, append-only tables — never by instructions to a model, and never by the model’s good behavior.

Zero LLM on the server

All reasoning happens in your own assistant. The server is deterministic — it cannot be talked out of its promises, because it cannot be talked to.

Your keys stay yours

Your integrations live in your assistant, where they already are. The rare key we must hold — a transcript service you connect — is encrypted per account, never readable back.

Isolation the database enforces

Row-level security with FORCE, tenant set per transaction, composite keys that make a cross-tenant reference unexpressible — proven by tests on every commit.

Memory that can defend itself

Append-only evidence, versioned beliefs, field-level provenance. Nothing is silently overwritten — beliefs are superseded, and the supersession is itself recorded.

Why it’s built this way

“The assistants everyone is building will keep getting more capable; the models guarantee that much. What the models do not guarantee is anything worth calling trust. Trust is architecture.”

Read the Pinolo Manifesto →

Give your working life a memory.

Sign in with your email — a magic link, no password to invent.