Taking GenAI for a Drive

We have seen multiple technology waves over the last 30 years:

• the Internet,
• mobile,
• cloud,
• social media,
• e-commerce,
• and data platforms.

Most of those technologies required specialists before ordinary users could meaningfully benefit from them.

Generative AI is different.

For perhaps the first time in technology history, the technology itself can teach you how to use it.

You can literally ask:

• “What can you do for me?”
• “How should I use you?”
• “What are your limitations?”
• “Help me learn prompting.”

And the technology responds directly to the user.

That changes everything.

This blog is a practical guide based on a recent session with CAs, investors, and professionals on how to approach GenAI without hype, fear, or unnecessary technical jargon.

# The Biggest Mistake People Make With AI

Most people approach AI in one of two wrong ways:

# Mistake 1 — Waiting to Understand Everything

People feel they must first understand:

• neural networks,
• transformers,
• LLMs,
• embeddings,
• vector databases,
• GPU architectures.

before they can start using AI.

That is unnecessary for most professionals.

You did not learn:

• internal combustion engineering before driving a car,
• TCP/IP before using the internet,
• or semiconductor physics before using a smartphone.

The same principle applies here.

# Mistake 2 — Trying It Once and Giving Up

Many people:

• ask one vague question,
• get a generic answer,
• conclude “AI is overrated,”
• and stop experimenting.

GenAI is not a search engine.

It behaves more like:

• a collaborator,
• an assistant,
• or a junior analyst.

The quality of output depends heavily on:

• the clarity of your thinking,
• the quality of your context,
• and iterative refinement.

# The Driver → Mechanic → Assembler Framework

This is the mindset I have personally used across multiple technology waves over the last 30 years.

It applies surprisingly well to GenAI.

# Step 1 — Start as a Driver

Most people should begin here.

The goal is simple:

use the technology before deeply understanding the internals.

Think about learning to drive a car.

Nobody starts by:

• opening the hood,
• studying engine mechanics,
• understanding fuel injection systems,
• or analyzing gearbox engineering.

You first learn:

• steering,
• acceleration,
• braking,
• parking,
• and road awareness.

GenAI should be approached the same way.

Start using it.

Experiment with it.

Drive it.

# Practical Advice

Use AI in areas such as:

• work,
• investing,
• writing,
• communication,
• summarization,
• learning,
• planning,
• brainstorming,
• and analysis.

Do not overthink the technology initially.

Focus on:

• capability,
• usefulness,
• and workflow.

# Step 2 — Drive in Different Contexts

This is where real learning begins.

Do not test AI only in one narrow scenario.

Use it:

• professionally,
• personally,
• analytically,
• creatively,
• in familiar domains,
• and unfamiliar domains.

# Most Important Advice

Test AI especially in areas where you already have expertise.

Why?

Because experts can:

• detect hallucinations,
• identify weak reasoning,
• understand nuance,
• and evaluate quality.

This helps you understand:

• where AI works extremely well,
• where it struggles,
• and where human judgment is still essential.

# Example Contexts to Explore

# For CAs

• explaining tax changes,
• drafting client communication,
• summarizing regulations,
• extracting insights from PDFs,
• comparing financial statements.

# For Investors

• company analysis,
• summarizing annual reports,
• identifying business risks,
• generating investment questions,
• comparing management commentary.

# For Professionals

• meeting summaries,
• first drafts,
• brainstorming,
• presentations,
• SOP generation,
• workflow acceleration.

# Step 3 — Become a Mechanic

Once comfortable using AI, you begin customizing it.

This is where concepts like:

• prompts,
• context,
• memory,
• and workflows

become important.

You are no longer just driving.

You are tuning the vehicle.

# The Three Important Controls of GenAI

Think of these as the steering wheel, brakes, and dashboard of AI systems.

# 1. Prompt

A prompt is the instruction or direction you give AI.

Bad prompt:

“How should I invest 25 lakhs?”

This is too generic.

Better prompt:

“I am a long-term Indian retail investor with moderate risk appetite and existing real estate exposure. How should I think about diversification across equity, debt, and gold?”

Notice the difference.

The second prompt contains:

• goals,
• constraints,
• context,
• and intent.

The result improves dramatically.

# Important Insight

Prompting is not about “magic words.”

It is about:

• clarity of thinking,
• structured communication,
• and iterative refinement.

Good prompts often reflect good thinking.

# 2. Context

Context defines the environment within which AI should operate.

Without context:

• AI gives generic answers.

With context:

• AI gives personalized and relevant responses.

Examples of context:

• profession,
• geography,
• risk appetite,
• business model,
• time horizon,
• communication style,
• audience type.

# Example

Instead of:

“Analyze ITC.”

Try:

“Analyze ITC from the perspective of a long-term retail investor focused on debt, promoter quality, and cash flow consistency.”

The difference is enormous.

# 3. Memory

Modern GenAI systems increasingly remember:

• preferences,
• recurring patterns,
• writing styles,
• and long-term interactions.

Over time, AI becomes more personalized.

This is where the system starts feeling less like:

• a search engine,

and more like:

• a collaborative assistant.

# Different AI Tools and Their Best Usage

Not all AI systems are optimized for the same tasks.

Tool	Best For
ChatGPT	reasoning, ideation, writing
Claude	long documents and analysis
Gemini	Google ecosystem integration
Perplexity	research with citations
Copilot	productivity workflows
Midjourney	image generation

One important lesson:

there is no single “best AI.”

Different tools work better for different use cases.

# What AI Is Surprisingly Good At

GenAI is already extremely useful for:

• summarization,
• communication,
• brainstorming,
• simplifying complexity,
• comparing documents,
• extracting insights,
• generating first drafts,
• structuring thoughts,
• and accelerating workflows.

# What AI Is Still Bad At

This is equally important.

AI still struggles with:

• factual consistency,
• legal accountability,
• nuanced financial judgment,
• deep strategic reasoning,
• and reliability across long workflows.

Most importantly:

AI sounds confident even when it is wrong.

That means:

• verification remains essential,
• human judgment still matters,
• and blind trust is dangerous.

# Demo Section

# Demo 1 — Prompt Evolution

# Starting Prompt

“How should I invest 25 lakhs?”

# Improved Prompt

(Add goals, constraints, time horizon, and risk profile)

# Advanced Prompt

(Add portfolio exposure, taxation considerations, and diversification requirements)

# Key Learning

Better prompts produce better outcomes.

# Demo 2 — Company Analysis

Example workflow:

• analyze annual report,
• summarize risks,
• identify red flags,
• compare management commentary,
• extract financial insights.

# Demo 3 — AI Memory and Personalization

Example:

• asking AI what it understands about the user,
• checking whether the context is accurate,
• refining memory over time.

Key insight:

Personalized AI becomes significantly more useful.

# Demo 4 — Failure Demonstration

Examples:

• hallucinated data,
• incorrect citations,
• fabricated numbers,
• overconfident answers.

This is one of the most important things to understand about GenAI.

# Key Lesson

Use AI:

• as an accelerator,
• as a thinking partner,
• as a workflow assistant,

but not:

• as an unquestionable authority.

# What Comes Next: AI Agents

Today:

• AI mostly answers questions.

Tomorrow:

• AI will increasingly execute workflows.

Examples:

• monitoring portfolios,
• drafting recurring reports,
• tracking compliance,
• reconciling invoices,
• generating research summaries,
• automating repetitive tasks.

This is where the industry is heading rapidly.

# Final Thoughts

The future is not about:

• humans versus AI.

It is about:

• humans working effectively with AI.

The professionals who benefit the most will likely be those who:

• experiment actively,
• learn continuously,
• verify critically,
• and combine domain expertise with AI leverage.

Do not fear the technology.

Do not blindly trust it either.

Take it for a drive.