Executive summary
I spent the better part of a decade in corporate engineering at large organisations before stepping back into entrepreneurship full-time, and what vanished overnight was not capability but support structure: delivery leads, product owners, analysts, stakeholders who used to share contextual load I had not noticed until I was on my own again. I had to learn whether I could stay oriented without a chief of staff and recover portfolio judgement after messy weeks.
The phrase second brain is jargon worth defining because it is not widely understood outside productivity circles (Tiago Forte's framing is the usual reference). I mean an external orientation layer where strategy and the relationships between ventures live outside my head, not a to-do list or an archive of everything I ever read. The goal is not to remember everything.
When several ventures share my attention, capture stays easy and portfolio reasoning stays hard. Without somewhere trustworthy to park that judgement, the exciting battle wins by default.
Week one I spent two or three days consolidating ten half-dead systems into one place. That has not literally paid revenue yet, but it rationalised overlap I could not reason about while everything was scattered. The information already existed; I was the dependency for the relationships between pieces. I can stay oriented across a portfolio without a chief of staff: memory structure that recovers from messy weeks.
Business insights
The apps that made capture easy were poor for reasoning across a portfolio. The bottleneck had shifted from generating plans to deciding what deserved trust and attention. I was rebuilding capability and judgement, not shopping for software. The value is reconnecting relationships between bets, not hoarding facts; the orientation layer has to earn its keep beyond search.
Insights
- When several ventures share one person's attention, the failure mode is strategic drift, not slow shipping.
- Capture tools optimise for intake; portfolio judgement requires relationships between bets to live outside your head.
- Consolidate until one place can answer how ventures connect; scattered systems make you the dependency for every link.
- Park strategy beside execution; a task list alone lets the loudest battle win by default.
- Hold the orientation layer to truths, judgements, and pointers; put task state and heavy artefacts elsewhere.
- Process capture asynchronously so you are not on call for your tools.
- Context rots; prune what no longer earns its keep (weekly review, human approval).
Hypothesis: When strategic relationships between ventures are externalised beside active work, portfolio alignment survives messy weeks because judgement is not re-inferred on every open.
What I observed: Week one consolidation made deferred longer-horizon work visible again; overlap I could not reason about while scattered became legible.
What I'd test next: Whether a weekly alignment check reduces productive-but-wrong weeks once the execution layer (Post 2) is in place.
How the orientation layer works
Consolidate before you optimise
The trigger was ten systems, all partially wrong, none trustworthy, and no single place that could answer how one venture connects to everything else. I adapted ACE (Atlas, Calendar, Efforts) from Nick Milo's Linking Your Thinking: reference that grows slowly, logs that compress over time, active efforts with a finish line. On top: Raw/ as the single inbox for phone dumps and messy captures, and _graveyard/ for archived notes with a reason and never deleted.
Raw/ → Atlas | Efforts | Calendar
→ _graveyard/ when done or obsolete
| Folder | Role |
|---|---|
| Atlas | Philosophy, methodology, people, assets |
| Calendar | Dailies, meetings, scorecards, agent logs |
| Efforts | Active ventures with a finish line |
Markdown on disk, IDE for work, git for history
Obsidian gave me markdown, links, and phone sync; I work in Cursor for the vault because that is where I already live for code. The vault lives in a private GitHub repo: boring add, commit, push, but it puts thinking under source control. If the system cannot match how I actually work, I end up maintaining two systems, which is the expensive failure mode.
Capture asynchronously, process without being on call
Phone captures land in Raw/. Ingestion classifies against the indexed vault, routes material into Atlas, Efforts, or Calendar, adds pointers where depth lives elsewhere, and clears the inbox; I deliberately avoided inline questions and phone pings because I do not want to be on call for my AI. The pain is not too many items; it is rebuilding context and inferring relationships on every open.
Junction box, not database
The vault is a junction box with a thin layer of useful context on top: instructions, truths, ideas, and pointers to depth elsewhere, not every report stored inline. Linear holds task state, WIP, and due dates; hosted docs and GitHub repos hold SEO reports from claude-seo, plus roadmaps and architecture diagrams rendered with Mermaid, material that would bloat context if inlined. Atlas is how I trade off in my own words, so agents inherit how I decide rather than a transcript of every meeting. A weekly janitor proposes merges and stale efforts; I approve. Nothing moves automatically. A digital twin of memory still has to prune like sleep: context rots, and depth must be cut so what remains stays useful.
What belongs in the vault
The orientation layer is what I described above: Atlas for judgement, Calendar for logs, Efforts for ventures with a finish line, Raw for capture, graveyard for retired context. Strategy stays with me; administration can delegate. Early signal from the first weeks: externalising those relationships kept longer-horizon ventures visible alongside tactical noise, including work I would otherwise defer indefinitely.
Conclusion
The problem, stated plainly: when several ventures share my attention and nobody else holds the thread between them, capture stays easy and portfolio reasoning stays hard. Strategic goals drift while exciting battles take the hours, and I become the dependency for relationships that used to live across a team.
The closest direction I found is an orientation layer that reconnects those relationships without trying to remember everything.
- Consolidate until one place can answer how ventures connect.
- Park strategic concerns beside the portfolio, not only on a task list.
- Hold that layer to truths, judgements, and pointers; put task state and heavy artefacts elsewhere.
- Process capture asynchronously so you are not on call for your tools.
- Prune what no longer earns its keep, and once a week check whether you stayed aligned with strategic goals or whether the loudest battle won.
That is the capability worth rebuilding: judgement and memory that survive when the support structure is gone.
More on me at rowanarcher.com.
Next: An Execution Layer That Survives a Solo Founder Week · All writing