A Second Brain for Your Business: Vision, Constraints, Decisions

Your business context lives in three places: old chat histories, note scraps, and your head. Every new AI session starts from zero, so you type the same paragraph again. Who the customer is. What you will not do. What you decided last month and why. A second brain for business is the fix: one place that holds vision, constraints, decisions and facts, and that a proposal engine can read before it suggests anything. In QuestDesk that place is called Brain. This guide covers what goes in it, how it turns records into next steps, and what to record first.

The re-explaining tax

The re-explaining tax is the time you spend restating your business before you can ask a useful question. It is paid in small coins: two minutes here, five there, and a worse answer whenever you skip it.

It has a second cost that is easier to miss. When the context lives in your head, each session gets a slightly different version. One day you say "evenings only", another day you forget to mention it, and the AI proposes a launch plan that needs a full week. You then argue with a plan that was never grounded in your real limits.

A company memory removes both costs. You record the constraint once, and every proposal after that is checked against it. Nothing in the record is a guess; it is what you wrote, and it stays until you change it.

Four record types that are enough

A business knowledge base for founders does not need a wiki with forty pages. Brain keeps four record types, and they cover almost everything you would otherwise re-explain.

Vision

Where the business is going and for whom. One or two sentences is enough to start: "A web app that turns long recordings into short clips, for solo creators who post weekly." You refine it later as you learn.

Constraint

What you will not do, or cannot do yet. Time, money, skills, ethics, scope. "Evenings and weekends only." "No paid ads until there are ten paying customers." "No features that require storing raw video longer than a week." Constraints are the records that most improve proposals, because they rule out whole categories of work.

Decision

Something you chose, with the date and the reason. "Chose a monthly plan over per-clip pricing, because creators want predictable cost." A decision log for startups is mostly this list. It stops you relitigating settled questions and gives any proposal a reason it can cite.

Fact

Something you learned that is true right now. "Three of five trial users export to vertical format only." "Average recording is forty minutes." Facts change; you update or retire them. Facts are what stop an AI from inventing numbers, which is the subject of the guide on hallucinated metrics.

When you record founder decisions and facts as separate types, you can see at a glance what was chosen and what was observed. That distinction matters more the longer the business runs.

From records to proposals

Brain reads the records and proposes next steps. Each proposal is a candidate quest: a short brief with a kind, a size and a reason drawn from the records. You open it in Control, edit it, and save it. Only then does it exist as a quest.

Take Clipwise (example): a web app that turns long recordings into short clips, run by one founder, evenings and weekends.

The founder records a constraint, "evenings and weekends only", and a fact, "three of five trial users only export vertical clips". The vision says weekly-posting solo creators. Brain proposes a quest of the "decide one number" kind: decide the maximum recording length for the free tier. The reason it gives points at the fact about vertical export and the constraint about time; a small pricing decision fits a weekend, a new export pipeline does not.

Without the constraint record, a generic assistant would happily propose "build a horizontal-and-vertical export engine this week". With it, the proposal is sized to the founder's real life. That is the difference between an AI that knows your business and one that knows businesses in general.

Proposals are not a queue that drains itself. Brain proposes; you pick. If you save none of them, nothing happens.

You edit, you save, you decide

Brain never starts work on its own. That is not a setting; it is how the thing is built.

Every proposal arrives as text you can change. You can reword the brief, shrink the scope, change the kind, or discard it. When you save, it becomes a quest in the inbox. Approving it, starting a worker run on it and accepting the result are three further, separate decisions, each of which needs your click.

This matters for a second brain for business in particular. The records are your intent. If a system could act on your intent without asking, a stale fact or a badly worded constraint would become a bad action. Because Brain only remembers and proposes, a wrong record produces at worst a wrong proposal, and you catch it at the edit step. The plan chapter shows how a goal and a proposed next quest stay visible together.

A useful status view reads those same records; it should not invent work or start a task.

Seeing the business as a Kingdom

Records are useful, but a plain list does not tell you where attention is needed. The Kingdom rooms make business work visible by station: planning, design, building, marketing, finance and the other parts of an online business. A station can have more than one agent, and the overview should tell you what is working and what needs you.

This is a view over the same business records as the practical Desk, not a second set of quests to maintain. When you return after a few days, the main goal and next quest should be easy to find. Open a station for detail, or go straight to the question or result waiting for your decision. The public 3D scene previews this direction; the roadmap states what invited members can use today.

Starting fresh versus bringing an existing business

The setup flow asks you to choose a path first. The two paths need different first records.

Starting fresh

Record the vision first, even if it is rough. Then record two or three constraints, especially time and money, because those shape the first quests more than anything else. Skip facts; you do not have many yet. A starter plan can propose customer discovery, a first useful offer and a result you can inspect. You edit the quests that fit your actual time and budget. The guide on turning a business idea into missions and quests walks through this path.

If you do not have an idea yet, the idea finder offers a fixed list matched to your strengths, reach, time and money. It is a list, not model output, so it will not invent a market for you.

Bringing an existing business

Choose "I already have a business". Record customers as facts: who they are, how many, what they pay. Record constraints as they actually are, including the ones you are slightly embarrassed by. Record what is working as facts, and record the decisions behind it, so the reasons are not lost. Then let Brain propose. The first proposals for an existing business are usually about protecting what works and testing one thing that does not, which is the right order.

For either path, the solo founder guide covers how the rest of the app fits around Brain.

What to record every week

A company memory decays if it is only written once. A short weekly habit keeps it useful.

• One fact you learned from a customer, a number or a quote you can attribute.
• One decision you made, with the reason, even if it was small.
• Any constraint that changed: more time, less money, a new rule.
• Retire any fact that is no longer true.
• Read the proposals, save the one or two that fit, discard the rest.

That is ten minutes. In return, every proposal for the following week is grounded in this week's reality, and you never type the introductory paragraph again.

If your business currently lives in scattered chats and your own head, join the waitlist. Once invited, record the vision and one constraint, and see what Brain proposes.