← Projects

QBO-A03 · Quobo Labs fleet

Quobo Labs Schedule

Labs applicationIn the lab

An AI chief of staff that interviews you: you talk, it files what you say into a self-correcting memory that writes your to-dos, plans its own next meetings, and briefs any fresh AI on your world.

Built and used daily inside the lab — the story is below.

Chapter one

The problem

Every productivity tool ever built assumes you will feed it. You will write the notes, file the tasks, update the statuses, remember to review. For a certain kind of person — the founder juggling companies, deals, a family, and forty open loops — that assumption is exactly the failure mode. The context lives in their head, it never gets written down, the task list rots, and every new AI they try starts from zero because it knows nothing about their world.

The product's own system prompt quotes its user verbatim: “I am not somebody who manages themselves well — I need a signpost as to what to do next and how to do it.” That person surfaces at 7am with a full day and no picture of what matters. The night before a high-stakes meeting, with the prep scattered across memory and email. Friday afternoon, staring at commitments they can no longer place. Ten minutes into a fresh AI chat, re-explaining who everyone is. Again.

The insight underneath Quobo Labs Schedule: these people can't write, but they can talk — and they will answer a good question at length when they'd never fill in a form. So this is software that asks instead of waits. It holds the meeting. The memory is a byproduct of showing up and talking.

The mechanics

How it works

  1. The system calls the meeting

    Press Start meeting and the AI reads your entire memory — plus open to-dos, recent debriefs, the questions it already asked, and the last three days of your inbox — and writes a short agenda tailored to your world. Or press Bring something up and talk first: a worry, a decision, something that changed. Either way, the meeting is with you.

    The dashboard — Start meeting, Bring something up, and the queue of meetings the system thinks should happen next
  2. You answer out loud, one question at a time

    Each question arrives with its reason stated — “Why I'm asking: …”. Press go and talk as long as you like; the browser transcribes live and pauses are fine. When you stop, a quiet pass checks proper nouns against the names it knows and fixes mishearings — only names, never your words. Drop files onto a question and they're read alongside your answer.

    A live question with its intent line, the transcript filling in grey as you speak
  3. Everything you say becomes structure

    Submit, and the AI files the durable facts into a weighted memory it calls the corpus, extracts to-dos with priorities and due dates, and flags anything that contradicts what you said before — the old entry is superseded in the open, strikethrough and all, never silently rewritten. A receipt tells you exactly what was captured.

    The capture receipt — “Captured 3 corpus entries and 2 to-dos. Corrected 1 outdated note. Queued 1 future meeting.”
  4. The corpus decides what happens next

    When the meeting ends, the system writes a debrief, folds what it learned into standing profiles of your people and projects, and reconciles its own queue of future meetings — mapping sessions, deep dives, prep before a real event, a debrief after it passes, an accountability sweep of what stalled. Each one carries its rationale. More corpus, sharper questions.

  5. One click briefs any other AI

    Every to-do can generate a context pack: a paste-ready brief that makes a fresh AI session instantly fluent in that one task — what done looks like, the background, what's still unknown, a glossary of the proper nouns, sensitive specifics flagged inline. Your context stops being trapped in the tool that gathered it.

  6. Done is judged by evidence

    Each to-do has a home page with a work log. Post a two-sentence update or a photo of the letter you finally got; the system files the facts back into the corpus, closes what's now resolved, and — when the evidence says the task is genuinely finished — offers a button: “This looks done — mark it complete.” It suggests. You confirm.

Field notes

Would this fit your life?

A founder-operator with everything in his head

The moment
7am, coffee in hand, two companies and a renovation in motion, and no clear picture of what today actually needs.
What they do
Starts the morning check-in and talks through it — rambling is fine. The system asks about the risks and deadlines he wouldn't have thought to write down, and skips what he skips without chasing.
The payoff
Fifteen spoken minutes and the day has structure: fresh to-dos, an updated memory, and tomorrow's questions already queued.

A fractional CFO with a high-stakes Thursday

The moment
Monday's check-in mentioned a bank negotiation, and the system quietly scheduled a prep meeting before the event — it's sitting in the queue with its rationale attached.
What they do
Runs the prep interview: what we know, what we still need, the game plan. Drops the term-sheet PDF onto an answer so the questions can press on what the document actually says.
The payoff
The debrief doubles as a walk-in brief, read in the elevator. After Thursday passes, the system converts the prep into a debrief meeting on its own.

Someone who starts brilliantly and finishes rarely

The moment
The Friday accountability sweep surfaces in the queue — the system is built to face stalled commitments directly, without nagging.
What they do
Talks through the open list: what moved, what stalled and why, what should be dropped. It even asks whether the plan they nodded along to two weeks ago was ever actually adopted.
The payoff
The graveyard of half-done commitments gets faced weekly, by a counterpart with perfect memory and no judgment — the one review they could never run on themselves.

A solo consultant with a 9pm worry

The moment
A decision is looping in her head — whether to part ways with her biggest client — and no meeting is scheduled for hours.
What they do
Presses Bring something up and talks for six unstructured minutes. The system files what's durable immediately, then opens a deep dive whose first question engages exactly what she raised, quoting her own numbers and people back to her.
The payoff
The 2am spiral becomes a working session. New priorities enter the system the moment they're felt, not when a scheduled meeting stumbles onto them.

No varnish

From the lab notebook

Where it stands

A working, opinionated system its builder runs his life through daily — single-tenant, gated to the lab's own accounts, and built in under three weeks of rapid commits. The full loop is real end-to-end: agenda, spoken answers, live extraction, debrief, self-scheduling queue. The internals are more careful than the timeline suggests — the post-meeting passes fail soft, a recovery route can rebuild everything from stored transcripts, and nothing in the memory is ever edited silently.

What's still rough

No tests of any kind. Single-user assumptions run deep — one inbox per deployment, prompts written about one specific person. The corpus page lists only the newest two hundred entries with no search or edit, and its weight meters render on the wrong scale, so most bars read full. Dictation needs Chrome or Edge; everyone else types. And every answer is a frontier-model call with a narrated ~30-second wait and real cost — this is not a cheap product to run.

What's next

No formal roadmap — only the direction the commits point: more channels feeding one memory, and more surfaces where the memory pays out. The name says schedule, and queued meetings already anchor to real-world event times that today only enter through conversation — wiring the queue to an actual calendar is the obvious next mile. Anyone beyond the founder using it would need per-user email auth first.

Next in the fleet

QBO-A04

Quobo Labs Education

Field study

A private AI workspace for special-ed paperwork: every draft is grounded in the student's own record, and nothing is written without the teacher's one-click approval.

Continue the tour →