Kunaal Naik

Directions Website

Kunaal Naik Hi, I’m Kunaal Naik—AI expert & automation strategist. I help professionals build AI agents using No Code & Python, even with zero tech background.

My goal: make AI simple, practical & powerful for everyone.

04/10/2026

I trained Claude to automate my job search.

Here are the 5 prompts I used 👇

1. LINKEDIN PROFILE OPTIMIZATION

Step 1:
Go to your LinkedIn Profile → Resources → Save to PDF.
Upload it to Claude.

Prompt:
“Audit my LinkedIn profile (attached) against senior [role] job descriptions at [target companies]. Find gaps in my headline, About section, and Experience including missing keywords, weak positioning, and buried accomplishments. Then rewrite my headline and About section to position me as a strong candidate optimized for recruiter search.”

2. JOB SCRAPING WITH APIFY

Prompt:
“Use the Apify MCP to scrape LinkedIn jobs for [role] posted within the last 7 days at companies with 50–5000 employees. Filter roles matching my background in [domain]. Return a ranked table containing company, role, location, salary, posting date, and a fit score (1–10) based on my resume. Flag the top 10 jobs I should apply to today.”

3. ATS RESUME OPTIMIZATION

Prompt:
“Here’s my resume and a job description. First, act as an ATS parser and extract every keyword, skill, and requirement the ATS will scan for. Then rewrite my resume bullets to naturally mirror that language while preserving my real accomplishments. Score the before and after ATS match rate and flag any keyword stuffing that feels unnatural.”

4. NETWORKING VIA CLAUDE CODE
Step 1:
LinkedIn → Settings → Data Privacy → Get a Copy of Your Data → Download.
Upload the export to Claude Code.

Prompt:
“Using my LinkedIn connections export, build a reusable networking workflow that takes a target company and returns:
1. My 1st-degree connections there
2. The best 2nd-degree paths with warm introductions
3. Personalized outreach message templates based on shared history
Save this as a reusable workflow I can run for any company.”

5. 1 POST PER DAY → 90 CONTENT IDEAS

Prompt:
“I’m targeting [role] at [companies]. Generate 90 LinkedIn post ideas across four content pillars:
1. Builder credibility
2. Lessons from my current role
3. Opinions on my domain
4. Behind-the-scenes of my job search
For each idea, include:
Hook,Angle,Format

30/09/2026

Comment “Claude” if you wanna learn to build AI agent for FREE!

Morning brief agent
It wakes up before you do, reads your calendar, your inbox and the news, and then hands you one page to start the day with.
Setup: Project + calendar and mail connectors

Meeting prep system
Before every call it pulls the company, the contacts and the recent news, then hands you three talking points and one risk to watch for.
Setup: Project + web search + a saved prep prompt

Inbox triage
It sorts what actually needs you, what needs a reply and what needs nothing, and then drafts the replies itself.
Setup: mail connector + rules in Project instructions

Prospect research pipeline
Give it a name and it hands back the full brief. Company, role, angle, opener, ready to send.
Setup: Project + web search + a fixed output template

The decision council
Four sub agents argue your decision from four different angles and then hand you the verdict.
Setup: one Project, four role prompts

Your second brain
Every note, doc and screenshot lives in one place, and you can just ask it questions in plain English.
Setup: Project + your files uploaded once

Content repurposing engine
Feed it one long piece and it hands back a week of posts, written in your actual voice.
Setup: Project + voice samples + format rules

The money system
Give it your income, spending and goals and it hands back a real plan, updated every month.
Setup: Project + your numbers + a monthly review prompt

Weekly review loop
Every Friday afternoon it tells you what moved, what stalled and what next week actually looks like.
Setup: a saved prompt and a recurring calendar block

Your voice clone
Feed it five things you’ve actually written and it stops sounding like a chatbot.
Setup: custom Style built from writing samples

Pick one and build it this weekend✴️

27/09/2026

Check link bio for FREE Masterclass on Agentic AI

Or Comment “AI” for link your DMs

20/09/2026

Before this if you wanna learn Agentic AI with me for FREE
LINK IN BIO 🙌🏻

How to use Claude Code like the creator (Boris Cherny)
1. Run parallel sessions (don’t overload one)
Boris runs 5 instances of Claude Code simultaneously in his terminal using 5 separate git checkouts of the same repo. He numbers his tabs 1-5 for easy reference and uses system notifications to know when any Claude needs input.

2. Invest in CLAUDE.md
The Claude Code team shares a single CLAUDE.md, checked into git. When Claude does something wrong, they add it to CLAUDE.md so it doesn’t happen again.

3. Start with Plan Mode
Shift+Tab twice to activate. Lock in the plan first, then switch to auto-accept. Claude usually one-shots ex*****on from there.

4. Codify every repeated workflow
Slash commands for anything you do many times a day. /commit-push-pr runs dozens of times daily. Subagents handle the rest: code cleanup, end-to-end testing, build validation.

5. Integrate external tools
Claude Code uses all of Boris’s tools autonomously:
- Searches and posts to Slack (via MCP server)
- Runs BigQuery queries with bq CLI
- Grabs error logs from Sentry
Tools don’t consume context until invoked. Config checked into git and shared with the team.

6. Long-running tasks
To ensure Claude can work uninterrupted, Boris uses a background agent to verify on completion. Agent Stop hook for deterministic checks, or the Ralph Wiggum plugin for autonomous loops.

7. PostToolUse hook to format code
While Claude generates well-formatted code 90% of the time, the hook catches edge cases to prevent CI failures.

8. Pre-approve safe commands
Skip the dangerous flag. Use /permissions to allowlist safe bash commands instead. Checked into .claude/settings.json so the whole team gets consistent behavior.

9. Opus 4.5 with Thinking for Everything
Slower and bigger than Sonnet, but you spend less time steering it and it’s better at tool use, so it ends up faster overall.

10. Ensure Claude can verify its work
If Claude has a feedback loop, it will 2-3x the quality of the final result. Verification varies for domains (Bash commands, test suites, simulators, browser testing, etc), so test what works

Most people are still using AI like a calculator.Ask a question.Get an answer.Do everything else manually.But that’s not...
17/09/2026

Most people are still using AI like a calculator.

Ask a question.
Get an answer.
Do everything else manually.

But that’s not where AI is heading.

Agentic AI doesn’t just respond. It acts.

Instead of asking:

“Write me a follow-up email.”

Imagine giving AI the goal:

“Find the lead → research the company → write the email → send it → follow up → update the CRM.”

That’s the shift from prompts to workflows.

For businesses, this could mean delegating repetitive work like lead follow-ups, customer support, invoice reminders, reporting, CRM updates and internal research to AI-powered systems.

The advantage won’t simply come from using more AI tools.

It’ll come from building better systems around AI.

If your business still has hours of repetitive manual work every week, there’s probably an opportunity to automate part of it.

Want to see where an AI agent could save hours in your business?

DM me “AGENT” for FREE MASTERCLASS link🙌🏻👇

17/09/2026

If you wanna learn Agentic AI…
Comment “Agentic” will send the FREE MASTERCLASS link in your DMs 🙌🏻

15/09/2026

Save this reel & Comment “Masterclass” to receive a FREE MASTERCLASS link where I will teach you 1:1 for free!

13/09/2026

Patrick Winston ran How to Speak at MIT for four decades, and his final 2019 lecture has over 22 million views on YouTube. He believed presentation is a skill, not a talent, something you can train the same way you train any other craft.

Prompt 1, opening: “ROLE: Presentation strategist trained in Patrick Winston’s MIT framework. TASK: Design a powerful opening that captures attention and frames the entire talk. STEPS: Ask for my topic, audience, and core idea. Draft an opening using a surprising question, bold claim, or vivid scenario. State the promise within 60 seconds. Cut all pleasantries. RULES: Never start with ‘Today I’ll be talking about…’ The first sentence must earn attention. No jokes, no gimmicks. OUTPUT: A 90-second opening script plus a rationale for each choice.”

Prompt 2, closing: “ROLE: Rhetoric architect using Winston’s Situation, Problem, Implications, Solution framework. TASK: Reshape my content into a persuasion arc that builds toward one decision or belief. STEPS: Review my outline. Map it onto the SPIS arc. Find the contribution moment. Write a closing that ends on strength. RULES: Every section must pull the next. The last slide is never ‘Thank You.’ It is your final statement of value. OUTPUT: Full structured outline plus a rewritten closing statement.”

Check link in bio for FREE Masterclass on Agentic AI!!

Meta just entered the AI Agent race — but it might already be late. 👀With Muse, Meta is making a move toward personal AI...
10/09/2026

Meta just entered the AI Agent race — but it might already be late. 👀

With Muse, Meta is making a move toward personal AI agents that don’t just answer questions… they can potentially understand context, use tools, execute workflows and take action on your behalf.

But open-source ecosystems are already moving fast.

So Meta’s biggest advantage may not be the technology.

It’s distribution.

Billions of people already use Instagram, WhatsApp, Facebook and Messenger. If AI agents become deeply integrated into these platforms, they could go from something only AI enthusiasts use to something almost everyone interacts with.

And that’s the bigger shift I’m watching:

AI that answers → AI that acts.

For founders, consultants and businesses, the question will soon change from:

❌ “Which AI tool should we use?”

to

✅ “What work can we delegate to AI?”

The companies that figure this out early could build a serious advantage.

What do you think — will Meta dominate personal AI agents, or will open source stay ahead? 👇

Want to learn how AI agents can actually work inside your business?

🎁 Join my FREE 90-minute AI Agents Masterclass.
🔗 Link in bio.

AIAutomation MetaAI FutureOfWork Entrepreneurship NoCode AIForBusiness GenerativeAI

09/09/2026

You just joined a new company. The codebase has 100+ microservices, Kafka, Redis, multiple databases, and external APIs everywhere. You have 10 minutes before someone assigns you your first production bug.

You have Claude Code. What do you do first?

Most people open the repo and say “analyze this codebase.” That’s like asking someone to summarize a city by looking at a map. You’ll get an answer but it won’t help you navigate.

The move is to make Claude onboard itself the way a senior engineer would on their first week. Not writing code, just building a mental model of how everything connects.

Here’s the prompt I’d use:

“You are a staff engineer joining this project for the first time. Your goal is not to write code. Your goal is to understand this codebase the way an experienced engineer would. Walk through the tech stack and framework versions, explain the overall system architecture, list the major services and what each one does, identify the entry points, trace one full request from the API to the database, map out every external integration (Kafka, Redis, queues, caches, third-party APIs), explain how config and secrets are managed, identify the main design patterns, summarize the coding standards and directory structure, recommend the 10 most important files a new dev should read first, flag any technical debt or risk areas, and produce a one-page onboarding summary. Do not modify any code.”

The reason this works is that Claude’s output is only as good as its understanding of your system. If you skip this step and jump straight to “fix this bug,” it’s guessing with no context. It’ll generate code that looks right but doesn’t fit how your team actually builds things.

After this runs you have a map of the entire system in minutes. Every prompt after that is more accurate because Claude already knows where things live, how data flows, and what patterns your team uses.

The quality of AI-generated code depends on the quality of the context you give it. Teach it your system before you ask it to change your system.

WANNA LEARN AGENTIC AI FOR FREE?
Check Link in Bio!

Address

HSR Layout
Bangalore
560068

Alerts

Be the first to know and let us send you an email when Kunaal Naik posts news and promotions. Your email address will not be used for any other purpose, and you can unsubscribe at any time.

Shortcuts

Share

Category