07:23Tristen O'Brien · Tutorial
The 6 Claude Slash Commands to Get You Ahead of 90% of People Using AI
A 7-minute field guide to the built-in shortcuts most Claude Code users never touch — and one custom command recipe worth stealing.
June 14thThe argument in one line.
Claude defaults to filling every gap with assumptions and producing everything in one pass — five disciplined habits override that default and eliminate the rewrite loop that wastes most knowledge workers' time with AI.
Read if. Skip if.
READ IF YOU ARE…
- You use Claude regularly for work deliverables and find yourself rewriting most of what it produces.
- You manage strategy documents, client proposals, board presentations, or reports and want outputs that survive stakeholder review.
- You lead or coordinate projects with multiple moving pieces and want to use AI to map dependencies rather than just draft copy.
- You are starting every Claude conversation from scratch and spending 5-10 minutes re-establishing context each time.
SKIP IF…
- You already use Claude Projects with loaded context and run multi-pass reviews — this covers foundational habits you likely have.
- You are looking for automation or API-level Claude usage; this is entirely conversational UI-focused.
The full version, fast.
The video argues that bad Claude output is almost always a context problem, not a model problem. Five techniques address the five places where context breaks down: before writing (Interrogation Method), after writing (Perspective Shift), during complex builds (Chain of Documents), at project kickoff (Breakdown Method), and across every conversation (Memory Stacking). Each is shown with the exact prompt used, a real output example, and a time estimate for how much overhead it adds — the highest-cost technique (Chain of Documents) adds about 20 minutes to any long-form deliverable.
Free for members
Chat with this breakdown — free.
Sign in and you get 23 free chat messages on us — ask for the hook, quote a framework, find the exact transcript moment, generate a markdown action plan. Bring your own key when you want unlimited.
Create a free account →Where the time goes.
00:00 – 00:47
01 · Hook + Promise
Promises five techniques that eliminate rewriting, sets up the contrast between typical one-shot prompting and what the five skills replace.
00:47 – 02:17
02 · Skill 1 — Interrogation Method
Tell Claude to interview you before it writes. Exact prompt shown. Claude asks 10-15 targeted questions covering audience, budget, Q2 results, success metrics. Output after interrogation is specific; without it, Claude assumes.
02:17 – 04:46
03 · Skill 2 — Perspective Shift
Run three consecutive review prompts post-draft: CFO (financial rigor), biggest client (relevance and generic language), skeptical board member (unvalidated assumptions, logical gaps). Each catches a different problem class. Under 10 minutes total.
04:46 – 06:54
04 · Skill 3 — Chain of Documents
Break multi-page deliverables into four validated phases: Brief, Research, Outline, Full Draft. Each phase in context before the next begins. Fixes brief in 2 minutes vs. rewriting 15 pages later.
06:54 – 08:56
05 · Skill 4 — Breakdown Method
Give Claude full project scope, ask for task decomposition with dependency, blocking, and parallel maps. Surfaces invisible chains (compliance review blocking launch). Output also serves as a VP status report.
08:56 – 11:02
06 · Skill 5 — Memory Stacking
Load Claude Projects with persistent context: role, company, stakeholders, templates, tone. Every conversation starts at full context. Weekly report goes from 20 minutes setup to a single paste. All prior skills compound with this foundation.
11:02 – 11:19
07 · CTA
Teases a 'complete playbook' for one-person operations with Claude, directs to a linked resource.
Lines worth screenshotting.
- Claude defaults to being helpful immediately, which means it fills every gap in your prompt with assumptions — the Interrogation Method forces it to ask instead.
- 10 to 15 questions give Claude more context than most team members receive in a full briefing meeting.
- Reviewing a deliverable from your own perspective misses every objection a CFO, client, or board member would catch in thirty seconds.
- Three perspective reviews take under ten minutes total and each one surfaces a completely different class of problem.
- The brief phase of Chain of Documents is the highest-leverage step: fixing a misdirected brief costs 2 minutes; rewriting a 15-page document costs days.
- A document produced through four validated phases reads like team output, not a single-pass draft.
- Urgency and importance are not the same thing — AI dependency mapping reveals which urgent tasks actually block everything downstream.
- A compliance review with a two-week turnaround discovered in week three slips a launch by ten days minimum; discovered on day one it costs nothing.
- Claude Projects make every prior technique faster because the model starts each session already knowing your business, stakeholders, and terminology.
- A weekly report that required 20 minutes of context-setting becomes a single paste and one prompt inside a loaded project.
- The difference between starting every Claude session from zero and loading persistent context is the difference between a new contractor and a six-month team member.
- The more specific you make the perspective reviewer, the more useful the feedback — a generic board member review is weaker than a named skeptical LP review.
Five places where Claude context breaks down — and how to fix each one.
WHAT TO LEARN
Most AI output problems are context problems: Claude assumes what it doesn't know, reviews nothing, and forgets everything between sessions — five named habits close each gap.
- When you ask Claude to write without giving it context, it fills every unknown with assumptions — asking it to interview you first with one question at a time forces specificity before any draft begins.
- Reading your own work from your own perspective misses every objection a CFO, client, or board member would catch immediately — three sequential stakeholder-perspective reviews take under ten minutes and surface categorically different problems.
- Single-pass generation of any multi-page deliverable produces a first draft, not a finished document — breaking the work into Brief, Research, Outline, and Full Draft phases catches structural and audience problems before they become 15-page rewrites.
- Starting a complex project with the most urgent task rather than the most critical one creates downstream bottlenecks — asking an AI to map dependencies, blockers, and parallel work streams before any work begins surfaces invisible scheduling constraints.
- Every Claude conversation that starts without loaded context wastes 5-10 minutes re-establishing who you are and what you do — a Claude Project loaded with your role, stakeholders, tone preferences, and recurring templates eliminates that overhead permanently.
- The five techniques compound: context loaded into a project makes the interrogation faster, the perspective reviews more accurate, the phased documents more relevant, and the dependency maps more precise because the model already knows your business.
Terms worth knowing.
- Interrogation Method
- A prompting habit where you instruct Claude to ask clarifying questions one at a time before producing any output, replacing its default behavior of filling gaps with assumptions.
- Perspective Shift
- Running multiple consecutive review prompts that ask Claude to evaluate its own output from the viewpoint of specific stakeholders — CFO, client, board member — each uncovering a different class of error.
- Chain of Documents
- Breaking a multi-page deliverable into four sequential phases — Brief, Research, Outline, Full Draft — where each phase is validated before the next begins, keeping all prior phases in the same conversation context.
- Breakdown Method
- Giving Claude the full scope of a complex project and asking it to decompose the work into tasks with explicit dependency, blocking, and parallelism relationships, surfacing non-obvious sequencing constraints.
- Memory Stacking
- Loading a Claude Project with persistent context — role, company, stakeholders, tone preferences, recurring templates — so every conversation inside the project starts with full business context already in place.
- Claude Projects
- A feature in Claude's interface that stores custom instructions and persistent context for a defined work domain, so the context carries across all conversations within that project.
Lines you could clip.
01:44
“The quality difference between assumed context and provided context is the difference between a draft you delete and a draft you send.”
10:05
“The new contractor needs a briefing every time. The team member just needs the task.”
06:10
“Reviewing an outline is 10 times faster than reviewing a full draft.”
Word for word.
Read-along
Don't just watch it. Burn it in.
See every word as it's spoken — crank it to 2× and still catch all of it. The same dual-channel trick behind Amazon's Kindle + Audible.
analogy
00:00Did you know there are five Claude techniques that can completely change the quality of what it produces for you? I've been using all five for the past year and they're the reason I stopped rewriting everything Claude gives me. So in this video, I'm going to show you each one with a real example so you can start using them today.
00:14The first skill is the one that changed my output quality more than anything else and it starts with telling Claude to stop writing. The typical approach is to open Claude. Type something like, write me a marketing strategy and Claude immediately produces two pages of generic recommendations that could apply to any company in any industry.
00:32The output is technically correct, but completely useless because Claude had zero context about your audience, your budget, your competitive landscape, what you've already tried, or what your CEO actually cares about seeing in a strategy document. The way you fix it is with a single instruction. Before Claude writes anything, tell it to interview you first.
00:49I'll type interview me about every aspect of this project before you produce anything. Ask one question at a time until you fully understand what I need. Next, Claude starts asking targeted questions and not surface level ones.
01:02It asks about your target audience and how they've shifted since last quarter. It asks what worked in q two and what didn't. It asks about budget constraints and whether you're optimizing for growth or efficiency, the who deliverable is for, and what format they expect, and what success looks like in specific measurable terms.
01:18By the time Claude has asked 10 to 15 questions, it has more context than most team members get in a briefing meeting, and the output it produces after that interrogation is a completely different quality than what you get from a single prompt. The recommendations are specific to your company, the metrics reference your actual targets, and the language matches the tone your leadership team expects.
01:37I used this on a client proposal last month that would normally take me three drafts to get right. After a 12 question interrogation, the first draft landed.
01:45The reason this works is that Claude defaults to being helpful immediately, which means it fills in every gap in your prompt with assumptions. The interrogation forces it to ask instead of assume, and the quality difference between assumed context and provided context is the difference between a draft you delete and a draft you send.
02:03This works on anything where the output needs to be tailored, Strategy documents, client proposals, board presentations, performance reviews, even email chains where the tone matters.
02:13Anytime you find yourself rewriting Claude's output to fit your specific situation, the interrogation would have prevented that rewrite. Once Claude has the full picture, the next skill makes sure the output actually holds up under scrutiny.
02:24The second skill is about catching problems before anyone else does and it works by having Claude review its own work from perspectives that aren't yours. The natural instinct is to read through Claude's output, make a few edits, and send it. The problem is that you're reviewing it from one angle, your own, but the people receiving it are reading it from a completely different one.
02:42Your CFO reads it for financial rigor, your client reads it for relevance to their problem, and your board reads it for strategic risk. A single review from your perspective misses what all three of them would catch instantly. After Claude produces any deliverable, I run three perspective reviews before I accept it.
02:58First, I'll type, review this as my CFO would. What financial gaps, unsupported claims, or missing ROI calculations would they flag?
03:06Claude immediately identifies things like vague cost projections, missing timelines on ROI, unquantified risk, and benefit claims that aren't tied to specific numbers.
03:15It catches the exact things a CFO would circle in red during a review meeting. Then I run a second pass. I'll type, now review this as our biggest client would, what feels generic, what doesn't address their specific situation, and what would make them feel like this was written for someone else.
03:29This review catches a completely different set of issues. The CFO review catches the numbers, and the client review catches everything that feels generic or irrelevant to their situation. It flags sections where examples don't match the client's industry, where the language is too internal, and where recommendations feel templated instead of tailored, and a third pass.
03:47I'll type, review this as a skeptical board member. What risks aren't addressed? What assumptions haven't been validated, and where are the logical gaps?
03:55The review from a board member's perspective gives a completely different result. It uncovered unaddressed risks, unvalidated assumptions, and logical gaps throughout the report, timelines that didn't account for legal review delays, and positioning claims a client could easily question based on how they use a competitor's product.
04:11A quarterly report I ran through these three perspectives last month caught issues I would have completely missed on my own. Each review takes about thirty seconds to run and two minutes to read, so the full three perspective review adds less than ten minutes to your workflow, and the output that survives three different perspectives is significantly stronger than anything a single review produces.
04:30You can also create custom perspectives based on whoever is actually receiving the work. If you're presenting to a technical team, review from their CTO's perspective. If you're pitching investors, review from a skeptical LP's perspective.
04:42The more specific you make the reviewer, the more useful the feedback becomes. The perspective shift improves finished work, but the next skill is about how you build complex deliverables in the first place. The third skill changes how you approach any deliverable that's longer than a single page.
04:56The default approach is to try to get Claude to produce a full document in one prompt. Write me a business proposal for expanding into the European market. Claude produces something that reads like a first draft because it is one.
05:07A single pass from a model that's trying to cover research, structure, tone, and detail all at once. Instead, you break the deliverable into phases where each phase builds on the last and gets validated before the next one begins. Phase one is the brief.
05:20I'll type, I need a business proposal for European expansion. Before we write anything, draft a one page brief covering the objective, the audience, the key questions we need to answer, and the structure of the final document. Claude produces a focused brief that becomes the blueprint for everything that follows.
05:35And this is where you spot problems early because if the brief is targeting the wrong audience or asking the wrong questions, you fix it in two minutes instead of rewriting a 15 page document later. Phase two is research. I'll type.
05:47Based on this brief, research the key questions we identified. For each one, give me the data points, the risks, and the opportunities. You get structured research for each question in the brief.
05:57If you have web search enabled, it pulls current data. If you've uploaded internal documents into the conversation, it references those as well. Phase three is the outline.
06:06I'll type, using the brief and the research, draft a detailed section by section outline with the key argument for each section and the supporting evidence from the research phase. The outline is where the structure of the final document take shape, and reviewing an outline is 10 times faster than reviewing a full draft.
06:22If a section is in the wrong place or an argument doesn't flow logically, you catch it here. And phase four is the full draft, which Claude writes with the brief, the research, and the outline all in context. The document that comes out of four phases reads like something a team produced over a week.
06:37The arguments build on each other, the data supports the claims, and the structure flows logically because each phase was validated before the next one started. I built a 15 page client proposal this way last quarter, and the partner reviewing it was convinced an entire team had worked on it. I've used this on proposals, strategy documents, board presentations, market analyses, competitive intelligence reports, and anything else where depth and structure matter more than speed.
07:00The phases add maybe twenty minutes to the total process, but the output quality is in a different league from one shot prompting. Chain of documents handles the quality of what you build, but it doesn't help when the project itself is massive and nobody knows what to start first. The fourth skill is for any project that's big enough to feel overwhelming, which in my experience is most of them.
07:18When a project lands on your desk, the temptation is to start working on whatever feels most urgent, but urgency and importance aren't the same thing. And starting with the wrong task can create bottlenecks that cost you weeks. Instead, I give Claude the full scope of a project and tell it to decompose it.
07:33I'll type, here's the full scope of our product launch. Break this down into independent tasks. For each task, tell me what it depends on, what it blocks, and what can run-in parallel.
07:42Claude identifies the dependencies that aren't obvious. The pricing page can't go live until legal reviews the terms, which can't happen until the pricing model is finalized, which depends on the competitive analysis. That chain of dependencies means the competitive analysis needs to start on day one, not day five when someone finally gets around to it.
08:00It also identifies tasks that can run-in parallel. While one team is working on the competitive analysis, another can be building the landing page template because those two work streams don't depend on each other.
08:10I used this on a product launch last quarter that had 14 moving pieces across three departments. Claude decomposed it into a sequence task list in about ninety seconds, and it flagged dependency chains that would have caused delays if we'd started in the order the team originally planned. One chain involved the compliance review, which had a two week turnaround that nobody had factored into discovered that dependency in week three instead of day one, the entire launch would have slipped by at least ten days.
08:37The breakdown also becomes a communication tool. When your VP asks for a status update, you can point to which tasks are complete, which are in progress, and which are blocked with a clear explanation of why. It turns a vague we're on track into a specific status report that builds confidence.
08:52The same approach applies to product launches, hiring pipelines, office relocations, system migrations, and any project where multiple people or teams need to coordinate their work without stepping on each other. The first four skills improve individual deliverables and projects, but the fifth skill is what makes all of them faster over time.
09:09The fifth skill is about eliminating the setup cost that slows down every conversation you have with Claude. If you're using Claude without projects or custom instructions, you're starting every conversation from zero. Every time you ask for something, you have to re explain your role, your company, your industry, your writing preferences, and the context of what you're working on.
09:27That setup takes five to ten minutes per conversation and over a week of daily use that's over an hour spent just re explaining who you are. The way around it is to set up a Claude project and load it with your persistent context. This includes your role and responsibilities, your company's products and market position, your preferred writing tone and format, any recurring templates or frameworks you use, the names and roles of key stakeholders you reference often, and your company's specific terminology or acronyms.
09:52Once that context is loaded, every conversation inside the project starts with Claude already knowing who you are, what you do, and how you like things done. The difference is like working with a new contractor versus someone who's been on your team for six months. The new contractor needs a briefing every time.
10:07The team member just needs the task. I set up a project for my weekly reporting workflow and a report that used take twenty minutes of context setting and back and forth now takes a single message. I paste in raw metrics and data, type weekly report, and Claude produces a draft in my format and tone using the exact metrics I always track because all that context is already loaded.
10:27I have separate projects for client communications, strategy documents, and internal memos, and each one carries its own context.
10:34So the output matches the audience without me specifying it every time. And memory stacking makes every other skill on this list faster. The interrogation method asks better questions because Claude already knows your business.
10:45The perspective shift reviews from more accurate angles because it knows your stakeholders by name. Chain of documents produces more relevant research because it knows your industry and competitive landscape, and the breakdown method identifies more accurate dependencies because it knows your team structure and approval processes.
11:02These five skills compound and make your process faster every time you use Claude. And if you wanna see how these skills fit into a full system for running a one person operation with Claude, I made a complete playbook that covers everything from daily workflows to client delivery right here. Thank you for watching, and I'll see you in the next one.
The Hook
The bait, then the rug-pull.
Five named techniques, exact prompts, and real output comparisons — the argument is that most bad Claude output is a context problem, not a model problem, and each technique plugs a specific place where context breaks down.
Named ideas worth stealing.
00:47concept
Interrogation Method
Prompt Claude to ask you questions one at a time before producing any output. Gets 10-15 targeted questions covering audience, constraints, success metrics, format expectations.
Steal forany deliverable where the output needs to be tailored to a specific company, audience, or stakeholder
02:17model
Perspective Shift
- CFO review
- Client review
- Board member review
Three consecutive post-draft review prompts from stakeholder viewpoints, each surfacing a different problem class.
Steal forany deliverable that will be reviewed by multiple stakeholders with different priorities
04:46model
Chain of Documents
- Brief
- Research
- Outline
- Full Draft
Four-phase validated build process for long-form deliverables. Each phase in context before the next starts.
Steal forproposals, strategy docs, board presentations, market analyses, competitive intelligence reports
06:54concept
Breakdown Method
Ask Claude to decompose a full project scope into tasks with dependency, blocking, and parallel relationships explicitly mapped.
Steal forproduct launches, hiring pipelines, system migrations, any multi-team coordinated project
08:56concept
Memory Stacking
Load Claude Projects with persistent business context so every session starts with full context already in place.
Steal forany recurring workflow — weekly reports, client comms, strategy docs, internal memos
How they asked for the click.
VERBAL ASK
11:02product
“if you wanna see how these skills fit into a full system for running a one person operation with Claude, I made a complete playbook that covers everything from daily workflows to client delivery right here”
Verbal pitch in final 17 seconds with implied link; no on-screen URL displayed
Visual structure at a glance.
