Your First Useful Claude Session.

Most first Claude sessions go the same way. You open the interface, type something vague, get a wall of generic text, and close the tab. You might try again a few days later. Eventually you conclude it's not for you — or not for your kind of work.

That conclusion is wrong, but it's also entirely understandable. Claude is genuinely useful for complex professional work. The problem is that first-session failure isn't about ability. It's about approach. Professionals with 20 or 30 years of domain expertise don't need tutorials about AI. They need a structured first session designed around real work they already do.

The short version: Your first productive Claude session isn't about learning a tool. It's about running one specific piece of your actual work through a structured three-part workflow — context setup, task execution, and review calibration. Senior professionals at The Leverage Years run this 90-minute protocol, complete one real work task using Claude, and finish with a calibrated prompt they can reuse. No technical background required. Professional judgment stays in the driver's seat throughout.

If you've been skeptical about whether Claude could add value to your specific kind of work, this session will answer that question from direct experience rather than speculation.

Who this is for

This workflow is built specifically for:

• Attorneys, partners, and GCs carrying drafting, research, or correspondence backlogs
• CPAs, controllers, and CFOs who need to turn data and notes into written communication
• Consultants and advisors producing client deliverables and status summaries weekly
• Wealth advisors writing recap letters, client updates, or meeting prep documents
• Senior executives and operators who manage significant written output across their week
• Professionals 40+ who have deep expertise but no patience for technical complexity

This is not for you if you are already running structured Claude workflows for professional work, or if you're looking for AI automation, integrations, or technical deployment. This is specifically the first session — for someone who has never gotten a genuinely useful output from Claude in a professional context.

---

Your 90-Minute First Session: Five Structured Steps

The goal of this session is simple: finish it with one real piece of work done and a prompt you can repeat. Not a demo. Not an experiment. Actual output you could use in your practice.

You need: a Claude account (claude.ai, free tier is fine for this session), 90 minutes without interruption, and one piece of real current work — a draft you're procrastinating on, a memo you need to write, a summary you've been avoiding, or a set of notes you need to turn into something.

Step 1: Identify the Right First Task (10 minutes)

The single biggest predictor of first-session success is choosing the right task. Most professionals choose the wrong one.

Wrong first tasks: anything requiring precise numbers you haven't verified, anything involving active litigation or live regulatory filings, anything requiring information only you possess, anything where you can't evaluate the output quality.

Right first tasks: a summary of a document you have in hand, a draft letter or memo based on facts you know well, a set of talking points from meeting notes you've taken, a routine client communication you write versions of repeatedly, a response to a known fact pattern.

The pattern is simple: choose something where you are the expert who can judge the quality of what comes back. This isn't a limitation — it's what makes professional use of Claude different from casual use. Your judgment is the quality filter. The session only works if you can run that filter.

Before you open Claude, write on paper: the task, the audience, the key facts or inputs, and what "good" looks like. This takes ten minutes and it's the most important ten minutes of the session.

Step 2: Build Your Context Frame (15 minutes)

The most common failure in professional Claude use isn't the prompt. It's the absence of context. Claude produces generic output when given a generic instruction. It produces specific, useful output when given a specific professional context.

Your context frame is a paragraph you write once at the start of every session. It tells Claude who you are, what kind of work you do, what your standards are, and what you're working on. Think of it as the briefing you'd give a highly capable new colleague on their first morning.

Write yours before you open Claude. A good context frame covers:

• Your role and the nature of your practice or business
• Your audience (who reads your work — clients, partners, board, regulators, counterparties)
• Your standards (what you mean by "clear," "professional," "appropriate")
• What you're not doing (what Claude should never assume or add)

An example frame for a mid-market M&A advisor might read: "I am a senior M&A advisor at a boutique firm. I work with privately held businesses in the $10M–$100M range, typically owner-operated. My audience is sophisticated — CEOs, owners, and their counsel. I write clearly and directly, without jargon and without padding. I am not a public company practice. Do not reference SEC filings, 8-Ks, or proxy-related matters."

Your frame will be different. The key is specificity. Vague frames produce generic output. Specific frames dramatically narrow the response space.

Step 3: Execute the Task With a Structured Prompt (30 minutes)

Once your context frame is pasted into the conversation, you're ready to give Claude the task. A structured prompt has four components:

Role: "You are acting as a professional drafter supporting a [your description]."

Task: A specific, bounded instruction — not "help me write a memo" but "draft a two-page executive summary of the attached meeting notes, organized under three headings: key decisions, open items, and recommended next steps. Use short paragraphs and direct declarative sentences."

Inputs: The actual material — paste the notes, the relevant facts, the existing draft, or the key points you want reflected.

Constraints: What Claude should not do. This is where professional context matters most. Include constraints around tone, format, claims it can't make, information it should not invent, and anything specific to your professional domain.

The format matters less than the completeness. What you're doing is narrowing the space of acceptable responses so Claude has to work within your professional standards rather than around them.

Run the prompt. Read the output. Do not expect perfection. What you're assessing is: did Claude understand the task, stay within the constraints, produce something in the right register, and give you a workable first draft?

If the answer is mostly yes, you're ahead. If the answer is partly no, that's useful information — proceed to step four.

Step 4: Calibrate With One Revision Round (20 minutes)

Revision is where most professionals either stop or improve dramatically. The key is to treat the first output as a rough draft from a colleague, not a final product and not a failed experiment.

Pick the single most important thing that's off. One thing. Give Claude a specific revision instruction — not "make this better" but "the second paragraph is too general. Rewrite it to be specific to the fact pattern I gave you in [reference to what you provided]."

This is calibration, not correction. You are teaching Claude — within this conversation — what your standards mean in practice. The revision round is also where you discover whether Claude can work at the level of detail your work requires.

A few principles for productive revision:

• Be specific about what's wrong, not just that something is wrong
• Reference the actual content: "the sentence beginning with 'Typically…' should be removed"
• If the structure is wrong, say so explicitly before asking for a rewrite: "The memo needs to lead with the recommendation, not the background. Reorganize it."
• If the register is off (too formal, too casual, wrong level of hedging), give an example of the right register — quote one or two sentences from your own past work

After one revision round, stop. You are not trying to perfect this document in the session. You are building a calibration log of what Claude got right, what it needed correction on, and what your specific prompt frame needs to say next time.

Step 5: Extract the Reusable Prompt (15 minutes)

This is the step most first-session instructions skip, and it's the one that determines whether you use Claude again next week or quietly stop.

At the end of your session, write a clean version of the prompt you'd use next time for this type of task. This is your reusable prompt — a template that encodes your context, your task structure, your constraints, and your key revision notes.

Store it somewhere accessible. A notes file, a document, anywhere you can pull it up next time. The value of this session is not just the output you produced. It's the working prompt that makes the next session of this type take fifteen minutes instead of ninety.

If you completed a client memo draft today, your reusable prompt is the template for every client memo of this type. If you completed an executive summary, that's your template. The session pays compound dividends if you extract the prompt. It pays very little if you don't.

---

Checklist: Before You Start the Session

Use this before opening Claude for the first time:

• I have identified one specific real task I can evaluate the quality of
• I have written down the task, audience, key inputs, and what "good" looks like — on paper
• I have written a context frame describing my role, practice, audience, and standards
• I have 90 uninterrupted minutes
• I understand that the first output is a draft, not a deliverable
• I have a place to save the reusable prompt when I'm done

You don't need to prepare anything else. You don't need a tutorial, a course on AI, or a background in technology. You need the task, the inputs, and a clear picture of your professional standards.

Where Professionals Over 40 Have the Advantage

There's a persistent myth that younger professionals adapt to AI tools faster. In practice, senior professionals have a structural advantage in productive Claude use that most observers miss.

The quality of Claude output scales with the quality of the professional judgment applied to it. That means depth of domain knowledge, the ability to recognize when an output is wrong or subtly off, the experience to write a specific constraint rather than a vague one, and the judgment to know when a draft needs more than revision — when the underlying structure is wrong.

These are capabilities that accumulate with experience. A first-year associate using Claude and a partner using Claude are not running the same tool. The partner has a sharper filter, better constraints, more specific inputs, and a more calibrated sense of what "good" means in context.

The 90-minute first session doesn't give you all of that — you already have it. What the session gives you is a structured way to apply it.

What Claude Can and Cannot Do in Professional Work

Use Claude for:

• First-draft production of documents you will review and revise
• Organizing and structuring information you've already gathered
• Summarizing documents when you hold the source and can verify accuracy
• Drafting routine communications you write variants of repeatedly
• Building out the structure of a presentation or memo you'll populate and edit

Do not use Claude for:

• Final deliverables without professional review (your review, not Claude's)
• Legal advice, financial advice, or regulatory guidance — Claude is a drafting tool, not a practicing professional
• Anything where the inputs contain unverified facts you haven't confirmed
• Generating data, statistics, or research you haven't independently sourced
• Client-facing communications that have not been reviewed against your firm's professional standards

The professional boundary is clear: Claude is a capable drafter and organizer. You are the responsible professional. That division of responsibility is not a limitation — it's what makes Claude safe to use in client-facing work.

How This Becomes a Repeatable Operating System

The 90-minute first session is a single workflow, but it contains the structure of every productive Claude session you'll run afterward. Context frame, structured prompt, bounded task, constrained output, review layer, reusable extraction.

That structure, applied to the two or three types of work you do most frequently, becomes a personal operating system. Most senior professionals have five to eight recurring work patterns that generate significant written output. A client letter, a variance commentary, a status report, a matter summary, a board memo. Once you have a calibrated prompt for each, your session time for those tasks drops dramatically.

The Leverage Starter course is designed specifically for this transition — from a first successful session to a set of calibrated, reusable prompts across your core work patterns. It covers the prompt vault, the review layer, and how to build the repeatable version of what you did in your first session.

Not sure which course fits your work? Take the 6-question course selector.

If you already have a sense of the workflow you want to build, the full course catalog maps each professional role to the right program.

Protecting Confidentiality in a First Session

A reasonable concern, worth addressing directly: what about client confidentiality?

For the first session, choose a task that does not involve confidential client information. This is not because Claude is insecure — it's because learning to use a new tool on sensitive material is bad professional practice regardless of the tool.

Once you have a working session under your belt, you can evaluate your firm's protocols for AI use, your jurisdiction's or professional body's guidance, and the appropriate context for using Claude on matters involving clients. That evaluation is a professional one, not a technical one, and it should involve your firm's leadership or general counsel.

For the first session, use a template task, a sanitized version of a real task, or a task involving your own practice information rather than client information. The workflow is identical. The calibration is identical. The output is yours.

Frequently asked questions

Do I need any technical background to run this first session?

No. The workflow requires a Claude account, a specific task, and the ability to assess professional writing quality. Both are things you already have. The 90-minute structure is designed for professionals with deep domain expertise and no interest in technical complexity. The technical overhead is minimal: an account, a browser, and a clear task.

Why 90 minutes? Can I do a shorter version?

The 90-minute frame reflects the realistic time needed to complete the five steps well: task selection, context framing, first prompt, one revision round, and prompt extraction. You can compress the session if your task is simple and your inputs are clear. The step you should not skip is Step 5 — extracting the reusable prompt — because that's what converts a one-time experiment into a repeatable workflow.

What if the first output from Claude is not good?

That's the most common first experience, and it's instructive rather than discouraging. A poor first output almost always signals one of three things: the task was too vague, the context frame was too thin, or the constraints were missing. Review your prompt against those three criteria before revising. The improvement from a sharper prompt is usually significant.

Is Claude appropriate for client-facing professional work?

Claude is a drafting tool. All client-facing work requires professional review before it reaches a client — this is true whether you drafted it yourself, had a colleague draft it, or used Claude. The professional standard doesn't change. Claude can accelerate first-draft production significantly; it does not replace the professional judgment that your clients are paying for.

How is this different from generic AI tutorials online?

Generic AI tutorials are designed for general users. This workflow is calibrated specifically for senior professionals who have deep domain expertise, high professional standards, and specific constraints around client confidentiality and output quality. The steps — particularly the context frame, the constraints layer, and the reusable prompt extraction — reflect the particular requirements of professional work, not consumer use cases.

After the first session, what's the natural next step?

Run the same workflow on a second task, ideally from a different work pattern. The second session should take significantly less time because your context frame is written and your approach is calibrated. After two or three sessions across different task types, you'll have a clear picture of where Claude adds the most leverage in your specific practice — and that's where the course-level investment starts to make sense.