I am writing this on a Tuesday in the first week of March, twenty-five weeks and roughly two hundred and sixty-five commits into building DuranteOS. Studio has been live for fifty-three days. The Hexagonal migration is nearly done. The Sentinel MCP server is the next-up build per last week's dispatch.
I have been deferring a knowledge-portfolio audit for two months. The audit feels like overhead until you do it, at which point it usually pays back inside a week. I am writing the audit in public this Tuesday because the discipline of the public version is what makes me actually do it, and because I want next year's founders to be able to read what mine looked like at week 25.
The single piece of professional advice that has compounded most for me in twenty years of writing software did not come from a software book. It came from chapter one of The Pragmatic Programmer (1999, expanded 2019), where Andy Hunt and Dave Thomas open with a metaphor: treat your knowledge and experience as an investment portfolio.
"As with financial portfolios, you must invest in your knowledge portfolio regularly. Even if it's just a small amount, the habit itself is as important as the sums."
The metaphor extends. Diversify across asset classes (different domains, different abstraction levels). Invest regularly (the cadence matters more than any single bet). Review and rebalance (skills decay, fields shift). Manage risk (do not put everything into one vertical). Accept that some bets will fail (and price the failure in advance).
I have lived inside this metaphor for two decades and never written down what my actual portfolio looks like. This post is that audit. It is honest about the bets I am making with my professional time, where I sit on each curve, and what I would change if I were starting the same year over.
This is also, more practically, the post I wish someone had handed me when I first considered the indie-founder path. The knowledge bets you make in your first year as an indie founder do not just shape what you ship that year. They shape the next decade.
The thesis in one sentence
Your knowledge portfolio is the only asset you fully control. Manage it deliberately — diversify, review, rebalance — and the compound returns over a decade will dwarf any single bet you make on a product or company.
The Pragmatic Programmers angle: the four investment rules
Hunt and Thomas's chapter offers four operating rules for a knowledge portfolio. Each maps directly to choices an indie founder makes about how to spend learning time.
| Investment rule (Hunt + Thomas) | Indie founder translation |
|---|---|
| Invest regularly (small amounts beat lump sums) | Block weekly time for learning outside the current product's needs |
| Diversify (correlated assets all decline together) | Don't bet only on one tech stack, one market, one customer profile |
| Manage risk (high-return bets need low-return ballast) | Pair experimental bets (new AI capabilities) with stable bets (database fundamentals) |
| Review and rebalance (skills decay; fields shift) | Quarterly audit: what skill have I let atrophy? What has emerged? |
The four rules are not orthogonal — they constrain each other. Investing regularly without diversifying produces deep specialization that becomes a single point of failure. Diversifying without managing risk produces a Jack-of-all-trades who is master of none. Etc. The discipline is in holding all four constraints simultaneously.
The Kent Beck angle: smallest experiment
Beck's Empirical Software Design substack adds a complementary discipline: every learning investment should be the smallest experiment that could produce signal. Reading a 600-page book is not the smallest experiment. Building a working prototype with one new capability is. Watching a tutorial video on a new framework is not the smallest experiment. Re-implementing one of your existing features in that framework is.
The combination — Hunt and Thomas's portfolio framing plus Beck's smallest-experiment principle — is what makes a knowledge portfolio work in practice. Without the portfolio framing, you over-invest in whatever is in front of you this week. Without the smallest-experiment principle, you over-invest in learning to learn (reading, watching, planning) instead of learning by doing.
My current portfolio, audited honestly
Here is what my actual knowledge portfolio looks like as of March 3, 2026, with Dreyfus-model proficiency estimates per the Pragmatic Pragmatic Thinking and Learning (2008) framework.
| Asset | Dreyfus level | Allocation | Trend | Why this allocation |
|---|---|---|---|---|
| Software architecture (general) | Expert | 25% | flat | Twenty years of foundation; maintenance investment |
| AI agent design (Claude Code, prompts, evals) | Proficient | 30% | growing | Core to DOS; deepest active learning |
| TypeScript / Next.js / React | Proficient | 10% | flat | Tooling, not destination; just enough |
| PostgreSQL / Prisma | Competent | 5% | flat | Adequate for current Studio needs |
| Distributed systems (event sourcing, CQRS) | Competent | 8% | growing | Studio's credit ledger; deepening |
| Brazilian SMB market dynamics | Competent | 7% | growing | Altyaa's wedge; needs more time |
| Direct sales / GTM | Advanced beginner | 5% | growing | Underinvested; the binding constraint on DOS |
| Content / writing in public | Proficient | 8% | flat | Builder's Compass + this blog series |
| Operations / finance / legal | Novice | 2% | flat | Outsourced; deliberate non-investment |
Three things to notice about this allocation.
First, the largest allocation is to a domain where I am only Proficient, not Expert. AI agent design is the area I am betting will compound most over the next decade. The investment is not "study the field" — it is "build DOS daily, ship Builder's Compass weekly to roughly 3,800 readers, accumulate the kind of judgment that turns Proficient into Expert." Beck's smallest-experiment frame governs every week of that work.
Second, direct sales / GTM is dramatically under-invested relative to its impact. This is the single biggest portfolio imbalance. The thesis (DOS) is technically sound; the execution (selling it) requires a skill I have explicitly under-invested in for two decades because as a consultant, sales came through referrals and reputation. As an indie founder selling a new category, that approach does not scale. Rebalance is overdue.
Third, operations / finance / legal stays at 2% on purpose. I outsource these and expect to continue outsourcing them. The opportunity cost of becoming an amateur accountant is higher than the cost of paying a professional one. Not every domain belongs in the portfolio; the ones that do are the ones whose returns compound for me specifically.
What "diversification" actually means at this stage
Naive diversification would say: spread across as many domains as possible. That is wrong for an indie founder. The right diversification is across correlated risk, not across count of skills.
Naive vs. correlated-risk diversification
- "Learn one new framework per quarter"
- "Read across 12 different domains"
- "Be a generalist who can talk to anyone"
- Result: surface knowledge across many areas; deep mastery of none; no compound returns because no single skill reaches the asymmetric-return zone
- Identify which assets fall together (e.g., "AI agents" and "TypeScript" are highly correlated; if AI tooling shifts, both become less valuable simultaneously)
- Hold deliberate uncorrelated counterweights (deep-systems knowledge, direct customer-development experience, persuasive writing)
- Concentrate within a thesis (AI agents) but hedge across the unknowns (will the substrate model shift? will the language change? will the market emerge in PT-BR or English first?)
- Result: meaningful depth in a coherent direction with explicit hedges against the most likely failure modes
The five questions I now ask before any new portfolio investment:
The five questions before adding a new asset to the portfolio
- What other assets is this correlated with? If everything I already hold falls together when this falls, the new investment adds variance without diversifying.
- What is the smallest experiment that would tell me whether this asset is worth deep investment? A spike, a prototype, a 2-hour talk with someone who knows it.
- What asset am I displacing? Time is finite. Adding an asset means subtracting time from another. Name what.
- Will I still hold this in five years? If no, why am I investing now? (Sometimes the answer is "yes, but only briefly" — that is fine if explicit.)
- Does this skill compound, or is it consumed? Learning a specific framework version is consumed (the framework changes). Learning why frameworks of this shape exist is compounding.
What public writing does to the portfolio
The single most underrated portfolio mechanic I have discovered is public writing as portfolio compounding. Writing Builder's Compass weekly for three years has done three things to my knowledge portfolio that no other investment has matched:
One. It forces me to consolidate what I am learning. Holding a half-formed intuition is cheap; writing the intuition down for nearly four thousand readers is not. The forcing function turns vague holdings into legible assets I can deploy in client work, in product decisions, in conversations with potential customers.
Two. It produces external feedback that grades my portfolio. When 40 readers reply to an issue with substantive critique, my hypothesis about an investment area is being stress-tested by the market. The feedback loop is faster and more honest than self-assessment.
Three. It attracts other operators with complementary holdings. The DOS founding cohort, the Altyaa beta users, the consulting referrals — almost all of them came through Builder's Compass readers who recognized me as someone whose portfolio overlapped with theirs. Public writing turned my portfolio into a signal rather than a private inventory.
The compounding mechanism
Public writing is the only portfolio investment I have found that compounds all four of the Pragmatic rules at once: it forces regular investment (weekly cadence), it diversifies (different topics each week), it rebalances by signal (you discover what readers respond to), and it manages risk (you discover quickly whether a thesis is wrong). Every indie founder I know who has compound returns on their work has some version of this loop running.
The three rebalances I am making this quarter
The honest output of an audit is what changes next, not just what currently exists. Three rebalances I am applying based on the March 3 portfolio above:
One. Move 5-7% of weekly time from "AI agent design" to "direct sales / GTM." The AI work is in a comfortable groove; the GTM gap is the binding constraint on DOS revenue. Smallest experiment: book three founder-customer development calls per week, every week, for the next quarter, regardless of how the calls go.
Two. Add a small new asset: "PT-BR business writing" (currently zero allocation). Altyaa's PT-BR captions are good; my own PT-BR writing voice is not. The rebalance: ship one Builder's Compass in Portuguese per month for a quarter and see what happens to engagement and reader composition.
Three. Stop investing in any new framework or tool that does not have a specific 6-month payback. The Studio stack is set; the DOS stack is set. Adding a new database, new ORM, or new framework version this year is portfolio drag, not investment. Discipline against shiny.
| Name | Type | Required | Default | Description |
|---|---|---|---|---|
| audit_cadence | quarterly | yes | every 90 days | How often to formally re-examine the portfolio. Less than quarterly produces drift; more than quarterly is overhead. |
| asset_count | 5-9 active | yes | 7 | Concentrating in 5-9 active assets prevents over-diversification while leaving room for hedges. |
| public_output | weekly | yes | ≥ 1 piece/week | Writing or shipping in public is the compounding mechanism. Weekly is the floor. |
| experiment_size | smallest possible | yes | 1 spike / 1 prototype / 1 conversation | Per Beck — never invest deeply in a new asset until the smallest experiment has produced signal. |
| outsource_threshold | opportunity cost calc | no | if outsource cost < 50% of opportunity cost, outsource | Operations, finance, legal almost always cross this threshold. Naming the threshold makes the choice deliberate. |
The compounding mechanism
Public writing as portfolio compound interest.
The big bet that frames it
The decision the portfolio is built around.
The GTM rebalance
Where the next quarter's investment goes.
Year retro (forthcoming)
The portfolio results, twelve months in.
The Pragmatic Programmer was published in 1999. The internet has changed three times since then. The frameworks I used in year one of my career do not exist anymore. The companies I worked for do not exist anymore. The single skill investment from that decade that still pays compound returns is the meta-skill of managing my own portfolio deliberately.
That is the durable asset. Everything else — the specific languages, the specific frameworks, the specific platforms — is consumable. The portfolio is permanent.
If you are deciding what to learn next, write down your portfolio first. The audit will tell you what to do without anyone needing to give you advice.
Was this page helpful?