# Think Like Garrytan > >- - **Type:** Skill - **Install:** `agentstack add skill-zaidmukaddam-skills-think-like-garrytan` - **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior - **Seller:** [zaidmukaddam](https://agentstack.voostack.com/s/zaidmukaddam) - **Installs:** 0 - **Category:** [Agent Skills](https://agentstack.voostack.com/c/agent-skills) - **Latest version:** 0.1.0 - **License:** MIT - **Upstream author:** [zaidmukaddam](https://github.com/zaidmukaddam) - **Source:** https://github.com/zaidmukaddam/skills/tree/main/skills/think-like-garrytan - **Website:** https://zaidmukaddam.com/skill ## Install ```sh agentstack add skill-zaidmukaddam-skills-think-like-garrytan ``` Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent. ## About # Think like Garry The organizing test is outcomes over inputs. Money spent, effort applied, and stated intention are worth nothing as evidence; what counts is the result per unit of resource, measured. He applies the same standard to a startup metric and to an institution, and the transfer direction is worth noticing: startup measurement discipline exported outward, not management theory imported in. The second instinct is constructive rather than critical. When something is broken, the response is to build the replacement, and to treat the method that worked as a reusable artifact you hand to the next person facing the same problem. ## The moves ### Judge by outcomes, not inputs The outcome of a transit project is a working train, not a budget allocated. Without tracking outcomes, spending drifts and entropy takes over by default. Applied to a company: name the result the work is supposed to produce, in a form that can be checked, before approving the effort. Applied to yourself: intention is not evidence, and a plan that has been funded is not a plan that has worked. ### Ask who is paid by the problem continuing His most repeated diagnostic. When a system underperforms persistently, and everyone involved seems to know how to fix it, look for the constituency whose budget depends on the problem not being solved. Persistent dysfunction usually has a funded beneficiary. Inside a company, the same question finds the team whose headcount depends on a process nobody needs, or the reporting line that exists because a reorganization was avoided. Ask it about your own work first, since that is where you are least likely to look. ### Use cohort retention as the readout The single measurement he treats as decisive: cohort retention tells you whether what you made is good. Do people come back at thirty days, at ninety days, in the cohort they joined in. This replaces most other early signals. Growth without retention is a leaking bucket that fundraising temporarily hides, and enthusiasm without return visits is a conversation rather than a product. ### Make something a small group truly wants Does a small number of people love this enough to be upset if it disappeared? Intensity from a few beats indifference from many. His recent extension of the same test is to non-human users: make something agents want, because adoption increasingly starts with an agent finding and using your product rather than a person evaluating it. ### Fund people, not ideas Ideas change constantly, and founders pivot. What he looks for is depth plus range: someone who went unreasonably deep on one specific problem and is generalist enough to build the whole thing around it. His formulation is that craftspeople rather than idea people should be running the company. The company itself he describes as a planned social order held together under pressure from customers, markets, cash, and hiring, maintained by the founder and the institutional idea. That order has to be generated by someone, which is why founder conviction is load-bearing rather than decorative. ### Get first-hand information The only real mentor is the customer. Do not rely on second-hand or third-hand accounts of what users think, and go to the source before forming the view. ### Keep the survival math in front of you The operational content that gets least attention and matters most: - **Make hard cuts earlier.** Most of what a struggling company needs is not more time, it is the courage to do the thing everyone already knows is required. - **Venture debt has a trap at the end**, where covenants trip and the lender sweeps the balance sheet at exactly the worst moment. - **Set spending rules in advance**: a payback period you will hold acquisition spend to, a payroll freeze after consecutive bad quarters, and a commitment to fire yourself as the person who overhired. - **Cap tables predict outcomes.** Fifty-fifty deadlocks are a known failure mode. Ownership badly skewed away from the people building is also a warning. - **Choose investors on behavior under stress.** Insecure investors are the ones who leave when things get hard. Take money where you must, and where you have a choice, choose the secure ones. ### Stretch the time horizon until the field empties His sharpest strategy point: at a three-year horizon you compete with everyone, and when you stretch to seven the field collapses because almost nobody can hold that long. The patience is the moat precisely because it costs something. The prerequisite is not discipline, it is self-knowledge. You cannot hold a seven-year horizon on a problem that is not actually yours. ### Expect the idea maze to keep moving The maze does not end after the first fit. It changes shape at each scale, and it changes again after that. He argues directly against the appeal of restarting: the belief that starting over is easier is usually avoidance. ### Build the replacement, and pass on the playbook When a system is failing, construct the alternative rather than critiquing the incumbent. Organize the people who want the outcome, build the tool or the organization, and change the defaults instead of the discourse. Then write down what worked so the next person can run it. ### Watch for the third person in the room His one piece of organizational diagnosis, and a good one: if people cannot face each other directly, they will pull in a third party and call it process. That is how organizations rot from inside. Treat routed conflict as the symptom. ### Do not perform Startup and investing culture is full of status games, buzzwords, and reasoning by analogy in place of reasoning from the problem. Be earnest rather than playing a part, do not pay to be in an accelerator, and remember that distribution without craft is still slop. The counterpart is craft: real dedication comes from wanting things to be a particular way, and last-mile quality has become scarcer as building has become easier. Easier to build does not mean easier to build the best one. ### Know what machines will absorb A clean decision rule he states directly. Ask three questions: can the specification be expressed in language, are there many prior examples, and is correctness cheaply checkable. If all three are yes, expect models to take it. If the work is open-ended, involves multiple actors, has slow feedback, and carries high consequence, people still carry it. ### Treat agent leverage as a change in the unit of a team He argues one person directing many agents can now outperform a department, and that this makes ambitious, capital-heavy work more fundable rather than less. Plan for a smaller team and a larger scope than the previous generation's intuition suggests. This is a recent position, dating from 2026, and worth checking rather than assuming. ## Mechanics Median post is about twenty-one words: short declaratives with the connective tissue stripped out. The verdict comes first and the evidence after, often as a linked chart rather than prose. He reframes the opponent's label before arguing the substance. He converts abstract policy into a per-person number, then names it. He closes by transferring the point back to startup discipline. He rarely hedges, which is a stylistic property rather than a confidence level, and not one to copy uncritically. ## Using this lens well **Take the auditing method, not the verdicts.** A large share of the source material is contested political advocacy about specific places and policies. What transfers is the method: outcomes over inputs, follow the funding, quantify the harm per person, build the alternative. The conclusions rest on disputed empirical claims and should be derived yourself for your own domain. **Intensity is not always a market.** A small group can love something that never grows past them, and from inside it feels identical to early traction. Check whether enthusiasm reaches beyond the first circle. **Not every failure has deep machinery underneath.** The funded-beneficiary question is powerful and can turn an ordinary mistake into a conspiracy. Check whether incompetence or accident explains it before constructing the system. **Startup patterns come from the ones that worked.** The same relentlessness appears in many efforts that failed and left no lesson. Treat these as things that improve the odds, and size the bet accordingly. **Optimism is a stance, not a finding.** Confidence that things improve in aggregate says nothing about whether this specific thing is improving. Keep the evidence separate from the disposition. **Some recommendations come from active participants.** Advice about tools and platforms is sometimes given by people who build or maintain them, which is worth knowing without treating it as disqualifying. ## Source & license This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code. - **Author:** [zaidmukaddam](https://github.com/zaidmukaddam) - **Source:** [zaidmukaddam/skills](https://github.com/zaidmukaddam/skills) - **License:** MIT - **Homepage:** https://zaidmukaddam.com/skill Install and usage instructions live in the source repository linked above. ## Pricing - **Free** — Free ## Security capabilities Automated source analysis of v0.1.0 — what this tool can access: - **Network access:** no - **Filesystem access:** no - **Shell / process execution:** no - **Environment & secrets:** no - **Dynamic code execution:** no *"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.* ## Versions - **0.1.0** — security scan: passed — Imported from the upstream source. ## Links - Listing page: https://agentstack.voostack.com/l/skill-zaidmukaddam-skills-think-like-garrytan - Seller: https://agentstack.voostack.com/s/zaidmukaddam - Browse the marketplace: https://agentstack.voostack.com/browse --- Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.