Turn Daily Inputs Into Durable Records
Your life already produces the raw material; the problem is that it disappears
You attended a meeting and three decisions were made. You replied to an email and promised to send a revised timeline by Friday. You read an article about pricing strategy and highlighted two passages. You decided to raise your rates. You received a warranty confirmation for a new appliance. Within 48 hours, most of these facts are gone from working memory. Within a month, the only trace is buried in email threads and calendar history you will never search.
Capture is the foundational second-brain operation. Every retrieval, every synthesis, every depends on records that exist because something was captured. without capture is an empty database. This chapter teaches five capture workflows that turn daily inputs into structured records with source attribution, confidence levels, and review dates.
Each project follows the same loop: raw input enters the system, the assistant extracts structured fields, marks the source and confidence of each field, and proposes a record for your review. You approve, correct, or discard. The approved record enters where every can use it.
Meeting capture turns a conversation into decisions, commitments, and contacts
After a meeting, you give the assistant your notes, a transcript, or even a rough summary of what happened. The assistant extracts: attendees, decisions made, commitments (who promised what by when), open questions, and follow-up items. Each extraction becomes a record in .
A single meeting often produces five to ten records across multiple types. A decision record for the scope change. A task record for the deliverable David committed to. An interaction record linking you to each attendee. A contact update for Sarah's new role. A project update for the revised Atlas timeline. One meeting, properly captured, feeds the morning brief, the task list, the relationship tracker, and the all at once.
The test for meeting capture is simple: does tomorrow's morning brief include a follow-up from today's meeting? If the brief says 'you owe Sarah the revised timeline by Friday,' the capture loop is working.
Email capture surfaces commitments hiding in your replies
Emails contain promises. 'I will send this by Friday.' 'Can you review this?' 'Let's revisit next week.' These commitments are scattered across threads and easy to forget. The email capture project scans recent email threads and proposes tasks and waiting-on items from the commitments it finds.
The assistant reads each thread and extracts: promises you made (tasks for you), promises others made to you (waiting-on items), deadlines mentioned, and action requests that have no response yet. It proposes records, and you approve or discard each one.
Email capture is the fastest way to prove the value of structured memory. The first time the assistant surfaces a you forgot you made, the capture project has justified itself. Run it once on a week of email, and you will find at least one promise that slipped through.
Reading capture keeps ideas from sinking into the saved-articles graveyard
You highlighted a passage about negotiation tactics. You saved a podcast summary about leadership transitions. You bookmarked an article on pricing strategy. Without capture, these become a pile of saved links you never revisit.
The reading capture project extracts structured records from your highlights: source title, author, date, the passage itself, your annotation if any, topic tags, and a resurfacing date. Each highlight becomes a record in , linked to related projects and topics. When you later ask 'what have I read about pricing?', the records return.
The key difference between saving and capturing: saving drops a link into a pile. Capturing creates a labeled record with source, topic, and a resurfacing schedule. The record lives where other modules can find it. Saved articles decay. Captured ideas compound.
Decision capture preserves the reasoning while the reasoning is still clear
You decide to take a new client. The reasoning is vivid right now: the revenue is good, the scope is manageable, and the timeline fits. In three months, if the project goes sideways, you will not remember why you said yes. Decision capture records the options, the choice, the reasoning, and a prediction you can review later.
The capture moment matters. When the reader says 'I think I am deciding whether to...' the assistant offers to create a decision record. It walks through: what are the options? What are you choosing? What do you predict will happen? What would make this decision wrong? When should you review it? The result is a record that future-you can learn from.
Household capture prevents warranty dates and model numbers from living in email
You buy a dishwasher. The confirmation email has the model number, warranty expiration, purchase date, and service provider. That email will be buried under thousands of messages within weeks. Household capture extracts these fields into a household record: item, model, warranty date, purchase date, store, and maintenance schedule.
When the dishwasher breaks in eighteen months, you ask the assistant: 'What do I know about the dishwasher?' The record returns: model number, warranty status (still valid for six months), the store where you bought it, and the service provider's contact information. One capture saves twenty minutes of searching through email when the appliance fails.
The same pattern works for home maintenance schedules. The assistant extracts filter replacement dates, annual service appointments, and lease renewal deadlines from confirmation emails. Each becomes a record with a reminder attached.