Hi, there!

Welcome to the 6th edition of Work in Beta.

In this edition, we tell you how to stop re-explaining yourself to AI every single time. We give you a step-by-step workflow to build your own voice profile once and use it forever.

We are also hosting our second AI WORKSHOP for non-technical professionals later this month on 21st February and would love to have you join.

So, let’s dive in!

If you are a non-technical professional and want to use AI at work without worry and in complete control, we are running a 2-hour session on the topic, limited to a small-group.

What you will learn:

If you wish to join the session, fill this form and we will reach out to you with details: https://forms.gle/D6TGEVoFBUqKPsP49

Here is a testimonial from the session we ran last week.

First time you use AI, your prompts look like this: "Summarize this." "Refine this." "Research this." Generic prompt, no context. The output is half-baked. You're not happy with it, but you don't know why.

Then you read somewhere that context matters. You start telling AI who you are, what you do, how you write, who your audience is. The output improves. Noticeably.

So you do it again. Every time you're copy-pasting the same 200-word block from an old chat into every new one.

At some point, this starts feeling like too much effort. So, you start skipping it again. The output drops. AI is back to being generic. You conclude, you could do this yourself in less time.

Sounds familiar?

Here's what we've seen across professionals who've tried AI seriously: this is where most people stop. Not because AI doesn't work. Because making AI work consistently takes too much manual effort every single time.

This isn't a prompting problem. It's a systems problem.

The question isn't "How do I write better prompts?" or "Why can't AI just remember me?"

The question is: How do you build a system where AI always has your context without you re-entering it every time?

The answer lies in a single structured file that captures who you are, how you work, and how you communicate. You build it once. You deploy it where AI can always access it. AI references it forever.

That's what the 5-step voice profile workflow below builds.

AI that knows your role, KPIs, and audience generates contextually sharp outputs.

These 20 questions capture everything AI needs to know about how you work, what you create, and how you communicate. Answer each honestly based on how you actually work, not how you think you should work.

WORK & EXPERTISE (5 questions):

WORK PROCESS & PREFERENCES (5 questions):

COMMUNICATION STYLE (5 questions):

CONTEXT & CONSTRAINTS (5 questions):

Time: 20 minutes. Answer directly and specifically. These answers become the "Work Context" section of your voice profile.

After you answer all 20 questions, copy AI's organized output. You'll need it for Step 3.

AI learns best from real examples which is a proven fact.

This step requires you to share actual work samples and explain what makes them effective (or ineffective). The goal is to give AI concrete reference points for quality, tone, and structure.

THE 5 QUESTIONS:

For the above questions, wherever applicable, paste the actual text or attach the relevant file/s.

Time: 15 minutes to gather examples + 5 minutes to run the synthesis prompt.

After you answer all 5 questions and AI provides the synthesis, copy AI's synthesis. You'll need it for Step 3.

Combine everything from Steps 1 and 2 and save as voice-profile.md or work-context.md - markdown works best for portability across AI tools.

This file is your complete voice profile. Everything AI needs to know about who you are, how you work, and how you communicate - in one document.

Time: 10 minutes to assemble and format.

Now make sure AI can access this file every time you interact with it.

There are five main deployment methods. Each has different persistence levels and works with different tools.

Method 1: File Attachment - Upload your profile at the start of each chat. Works everywhere. Best for testing new tools or occasional use.

Method 2: Projects - Create a workspace where your profile is always loaded. Works in ChatGPT (paid) and Claude (all plans). Best for distinct work contexts.

Method 3: Custom Assistants - Build a dedicated assistant with your profile baked in. Works in ChatGPT (Custom GPTs) and Gemini (Gems). Best for repetitive work types.

Method 4: Skills - Auto-loads your profile in every conversation. Claude only. Best for default voice across all chats.

Method 5: AI Memory - Ask AI to store your profile in long-term memory. ChatGPT only. Best for persistent context without manual steps.

Pick the method that matches your workflow. Most people use a combination: Projects for client work, Skills for default voice, file attachment for new tools.

Your first draft won't be perfect. That's expected.

Generate content using your profile. Read it out loud. If anything sounds off (wrong tone, wrong phrasing, wrong structure) update the profile with explicit rules about what to avoid or emphasize.

Three maintenance triggers:

First 5 uses: Expect 3-5 refinements. Track what feels off and fix it immediately.

Every 2-3 weeks: Review your profile against recent writing. Update it to match how you actually write now.

When facts change: New job, new role, new responsibilities - update the file the same day.

Your voice profile is a living document. It evolves as you evolve.

Mistake 1: Skipping the 20 questions and going straight to style Work context is 60% of the value. Without knowing your role, audience, and KPIs, AI writes beautifully but for the wrong audience.

Mistake 2: Describing your work instead of showing examples "I write professionally" tells AI nothing. Sharing 3 examples of your actual work gives AI concrete patterns to match.

Mistake 3: Vague style descriptions like "professional and helpful" Be specific: "Conversational but direct. Max 2-sentence paragraphs. Always open with question."

Mistake 4: Building once, never updating Update every 2-3 weeks. Update immediately when facts change (new role, promotion, restructure).

Mistake 5: Using the same voice for every context Your voice profile is your default. Board presentation vs Slack message vs client proposal - these need different voices. Build 2-3 profiles if you switch contexts often.

Mistake 6: Skipping the synthesis prompt in Step 2 Without AI's pattern analysis, you miss 60% of what makes your writing distinctive. Let AI find patterns you don't consciously know.

Day 1: Answer the 20 questions. 20 minutes. Answer honestly - how you actually work, not how you think you should. Run the prompt from Step 1.

Day 2: Answer the 5 questions along with examples of your work. Upload to ChatGPT or Claude. Run the writing analysis prompt from Step 2.

Day 3: Combine the analysis from Day 1 & 2. Save it as voice-profile.md.

Day 4: Deploy it. Pick one method from Step 4. Test with 3 different requests.

Day 5-7: Refine. Read outputs out loud. Update your profile when something sounds off.

By Day 7, you'll have a working voice profile. It won't be perfect - but it'll save you 10+ minutes every time you use AI.

Most people use AI. Few have AI systems.

The difference is persistence - not better prompts, not smarter tools, but a system that makes your context available every single time without extra effort.

Build your voice profile once. Stop explaining yourself forever.

Thanks for taking 10 minutes to read through this newsletter. Please do let us know what did you think of it especially on how useful was it for you.

You can also reply to this newsletter email if you want to share something specific with us. Have a great week ahead and see you again next week!

-Sonali & PD