How AI is Supercharging Our Workflow!

By Shuva (Senior Product Leader) and Gandharv (EX-Founder/Startup Consultant): Lately, we've turned AI models into our extended team of analysts, designers, and associates. Especially for getting a quick start on a 0-1 problem or simple optimizations. Here's our practical guide for product & growth managers. We are writing this on Linkedin (and we all know Linkedin is OnlyFans for Middle Managers), so this is suited for fellow middle managers. For deep wizardry and full spectrum builds - Twitter threads are better.??

AI Tools: Your Digital Dream Team

Most LLM-based chat applications converge on similar core capabilities but possess significant nuances (“same same but different”). Claude excels in prototyping, while Perplexity excels in research. However, they have quirks—even paid versions time out, and free ones can be unreliable - bailing out mid-conversation. Our solution? Combine high-end tools (GPT-4) for complex strategy and task breakdown with dependable free ones for the sub-tasks.

Wait, What Do We Actually Do with AI?

Whether we're “soaring at 30,000 feet” or "burning rubber on the ground" (or occasionally hitting rock bottom and digging further), our work demands quick shifts between altitudes. AI models have become adaptable teammates who can handle both worlds and give an instant head-start. Here's how we put it to work.

1. Strategy & Analytics (AKA The Big, Fuzzy Stuff)

POV Creation/ Ambiguous Questions (Market Sizing, What-Ifs, Strengths and Weaknesses): Is this problem big and ugly enough to wrestle with? What does the product roadmap look like? Which growth levers are we yanking? Ideal org structures? Competitive strategies? So many questions, so little caffeine.

  • Example: “Market Sizing of lead qualifying AI receptionist service for Indian businesses”
  • Suggested Application: DeepSeek R1, ChatGPT-01
  • Sample: Video and PDF
  • Remarks: Quick on the draw, though sometimes takes a coffee break without telling us. Doesn't provide clickable links-you know, keeps us on our toes.

Brainstorm Product Roadmaps, Growth Levers, Org Structures, and Competitive Strategies:

  • Example: “Give me potential Growth levers across Marketing, Product and Org for India’s largest Jobs platform ”
  • Suggested Application: DeepSeek R1, ChatGPT-01
  • Example: Video and PDF
  • Remarks: Our go-to brainstorming buddy-never judges our wild ideas. Might pass them on to Uncle Sam though.

2. Research and Getting Management On Board

Reports with Citations. Find Niche Data but Cross-Check Sources

  • Example: “Find "ARPU Trends in EdTech”
  • Suggested Application: Perplexity
  • Example: Video?
  • Remarks: It's like having a librarian who whispers sweet data points into your ear. Gives sources, links, and even suggests follow-up questions. Overachiever much? But Do verify the links. Occasionally, Perplexity sees data in links that isn't there, but then we’ve all done that, haven’t we??

Answer Side Questions (e.g., "ABM Org Structure Examples") while Drafting Plans:

  • Example: “If I have to kickstart ABM, what kind of org structure am I looking at”.?
  • Suggested Application: Bing (with Think Deeply)
  • Example: Video?
  • Remarks: Reasonably quick, free, and unlimited. It's like having a helpful intern who's powered by coffee and optimism.

Parse 100-Page Reports for Key Insights in Seconds:

  • Example: “Go through Udemy’s annual report and give me enterprise vs. Consumer revenue split”?
  • Suggested Application: Perplexity (Upload PDFs)
  • Remarks: Other LLMs don’t let you do this-slackers. Who needs speed reading when you've got AI?

Go Through Detailed Plans and Critique:

  • Example: “In my detailed report for my CXO, identify inaccuracies, weaknesses and missing points”.?
  • Suggested Application: Gemini
  • Remarks: Lets you paste in 500+ lines and gives you the lowdown.

3. Data Analysis & Decision Making

Crunching Numbers and Simulating Scenarios: Analyze data, play out "what-if" scenarios, and pressure-test decisions.?

  • Example: “Where's our traffic coming from? What's our Customer Acquisition Cost (CAC) across multiple segments? How does it change when we throw money at ads like confetti at a parade?”?
  • Suggested Application: ChatGPT-01
  • Example: PDF
  • Remarks: Our crystal ball for business decisions-without the mysterious fortune-teller fee.

4. Pictures & Media

Visual Wizards: Create posters for customer collateral in the blink of an eye, and whip up images for landing pages. Because a picture is worth a thousand words or at least a few clicks.

  • Example: “Make an AI Receptionist Image for my Brochure”.?
  • Suggested Application: WhatsApp (Meta) AI, Canva AI
  • Example: Video
  • Remarks: Fast, unlimited image generation from your phone or desktop with Whatsapp. Canva AI needs paid plans, but hey, who said magic comes free?

5. Product Ideation

Rapid UI prototyping: Ideas to interactive mockups in minutes → much quicker idea diffusion and alignments?

  • Example: “Create simple App screens for a new Shoe-Spa business”
  • Suggested Application: Claude
  • Sample: Shoe Spa App prototype and Sample Prompts
  • Pro-tip: Hit a timeout on long chats - esp. in the final stages of refining? Just start a new chat, paste the image, ask Claude to recreate the design pixel-perfect, and restart

More on how AI models are enabling a few native use-cases inside the product for a later post.?

6. Landing Pages (& Simple Workflow Websites)

Build, test, and launch - perfect for rapid market validation and user feedback.

  • Example: “Create simple landing page for a new Shoe-Spa business for testing demand using google ads”
  • Suggested Application: lovable.dev?
  • Sample: Shoe Spa App prototype

Closing

  1. Never trust, always verify: Cross-check AI outputs (e.g., Perplexity citations + Bing).
  2. Prototype fast, iterate faster: Use Lovable/ChatGPT for 1-hour MVPs.
  3. NEVER commit very aggressive timelines: Prototype speed is misleading. Production AI needs a high tuning time for accuracy, design polish, and scale.
  4. Let AI disagree: Tools like 01 expose blindspots (“Did you consider CAC payback?”).

And there - That’s our ode to AI (may the AI overlords forever be pleased with us)! Of course, for all the micro-manager out there, who now think they’ve finally found a tool that listens to their last instruction, you’re in for a surprise! You’ll realize you can’t micromanage AI either.?

Such an interesting read! Using multiple LLM tools for different tasks really shows how AI is shaping daily workflows. Which tool has surprised you the most with its capabilities??

回复
Rahul Chadha

Building Apna | Ex-BYJU'S | Growth & Strategy Professional

1 个月

Well-articulated read!!

Aravind Balagi Prasad

Building Ather Energy ! Head of Charging Infrastructure Business

1 个月

Great write up Shuvajyoti Ghosh and Gandharv Bakshi !

Utkarsh Goklani

Storyteller | Business Leader | IIMA | BITS

1 个月

Lovely article. Currently, I find AI to be a highly overconfident teen - who quotes with confidence first, and seeks forgiveness later when caught fabricating data points ?? Have you guys seen any major performance differences between Claude Sonnet and Haiku? Solid work btw Shuvajyoti Ghosh Gandharv Bakshi

Devilal Sharma

Founder - Famdo | Product & Tech Leader | CTO | AI Toilet for Preventive Health @ Medic.Life | Ex-Broadcom, Ex-Microchip | IIT Madras

1 个月

How is your experience with LLMs on math logics? For me it still solve math logic badly and can’t be trusted at all; sometimes need to write a code and verify that math result - even the o3-mini-high.

回复

要查看或添加评论,请登录

Gandharv Bakshi的更多文章

其他会员也浏览了