Building a successful niche product as a wedge into a bigger market

Neel Bhattacharya saw a problem and new AI would be the solution. Then, she waited years for voice AI to get to where it needed to be for her idea to succeed — and in 2023, she finally built TheSolarAI.

Now, it's making $8k MRR serving as a wedge into its horizontal counterpart, LeadTrackAI.

Here's Neel on how she's doing it. 👇

Starting with a wedge

I'm not building an "AI company." I'm solving the problem I've obsessed over for 20 years: Why do businesses lose customers they've already paid to acquire?

My background is enterprise marketing automation — IBM Unica (now IBM Marketing Cloud), Salesforce, Adobe Campaign. I built multi-channel orchestration systems before "marketing automation" was even a buzzword.

During that time, I saw a pattern: Companies that responded fastest across voice + SMS + email won. Not the biggest budgets. The fastest, most coordinated responses.

Then, I moved into Australia's energy sector. Solar companies were spending $50K+ monthly on Google Ads, generating hundreds of leads, and taking 4+ hours to call them back.

Buyers choose whoever responds first.

By 2023, AI voice technology finally matured enough to solve this. So I built TheSolarAI.com — a voice + SMS + email orchestration system that responds to solar leads in under 60 seconds and converts them into appointments.

Why voice? Because $20-50k decisions need human conversations. But businesses can't staff 24/7. AI bridges that gap.

We're currently serving enterprise clients across Australia's solar sector with TheSolarAI.com, and we're expanding the model to LeadTrackAI with automotive, real estate, home services — just about any industry where speed + multi-channel follow-up = revenue.

Solar was the wedge. LeadTrackAI is the horizontal platform.

Performance pricing

We're at about $8k MRR. We use an interesting business model — it's 100% performance-based.

Clients pay $0 upfront. We only earn commission when leads convert to appointments. Why? After 20 years implementing enterprise platforms, I learned that CFOs buy outcomes, not features.

"What if it doesn't work?" was always the objection I heard, so I removed it. Pay nothing until it works.

Yes, it's capital-intensive. Yes, it creates lumpy revenue. But it aligns incentives perfectly. We're partners, not vendors.

Here are the metrics the important metrics:

• 18-month average client retention

• 30-90% conversion rate improvements for clients

• Zero churn in the past 5 months

Performance pricing creates evangelical customers. That's worth more than predictable MRR.

Pricing issues

That said, performance pricing nearly killed the business.

I pay AI/SMS/infrastructure costs upfront, then wait 30-90 days for commission. So, in the early months, we were gaining clients and getting them strong conversions, but bleeding cash while we waited for payment cycles.

If I restarted, I'd probably rethink pricing in the beginning and go hybrid instead. Innovation in pricing is great, but you need to be able to afford your business model.

Building workflows

The AI was easy. The workflow was hard.

After 20 years building enterprise marketing systems, I knew:

• Which touchpoints mattered (voice → SMS → email)

• When to follow up (60 seconds, 2 hours, next day, 3 days, 1 week)

• What to say at each stage (qualify → educate → create urgency → close)

I started by Frankensteining an MVP using GPT-4, Twilio, a VPS, and Google Sheets as my database (yes, really). Ugly, but functional.

Three weeks in, I tested on 50 "dead" leads a Brisbane installer had abandoned. Result: 46% contact rate, multiple booked appointments, closed deals. He asked to use it immediately. That was validation.

It took me two months to build the real platform. Then it took me six weeks to get my first paying customers.

I wasn't inventing technology. I just was applying 20 years of enterprise expertise to AI capabilities that finally caught up to the strategy I'd known would work.

Challenges with voice AI

Voice AI quality was inconsistent in the beginning. It would talk over customers, mishear Australian accents ("solar" → "sola"), fumble objections, etc.

It took four months of fine-tuning conversation flows, accent training, and objection libraries.

Customers helped — they knew what "good" sounded like from analyzing millions of conversations over the years.

Reliability over hype

Reliability scales better than hype. So, when it comes to my stack, here's my motto: Use proven tools; innovate in the application.

That's why LeadTrackAI (and our earlier platform, TheSolarAI) runs on a deliberately “boring” but powerful stack.

We use:

• OpenAI GPT-4 Turbo for logic

• ElevenLabs for natural voice

• Deepgram for real-time transcription

• Twilio for calls and SMS

• The backend runs on Python/FastAPI, PostgreSQL, and Redis, hosted on AWS for speed and scale.

• Next.js/React powers the dashboard, and we integrate directly with Salesforce, HubSpot, and Pipedrive for instant adoption.

The result: Scalable voice automation handling 1,200+ daily interactions.

Because in the end, technology is 20% of success — understanding the workflow is 80%. That’s our real competitive edge.

Growth via relationships

To grow, I leveraged years of relationships and credibility.

Most founders talk "growth hacks." My strategy: Call people I've known for 15 years and say, "I built something that solves the problem we've been complaining about forever."

Most founders can't do this because they're building for industries they discovered last quarter. Your network IS your moat when you've worked in your industry for years.

Here's the breakdown:

1. Direct network

• 50 personalized LinkedIn messages to former colleagues, consultants, conference connections

• 30% of demos → paying clients

2. LinkedIn content

• Behind-the-scenes building, client wins, industry insights

• Positioned as expert/educator, not vendor

3. Industry events

• All-Energy Australia, Smart Energy Council events

• Shared frameworks, never mentioned product

And here's what DIDN'T work:

• ❌ Google Ads (expensive, wrong audience)

• ❌ Facebook Ads (B2B solar doesn't browse FB for SaaS)

• ❌ Cold email (low response, felt spammy)

If there was a do-over

If I could do it again, here's what I'd do differently:

• Hybrid pricing sooner

• Build leadership team by month 3 and get the right team to complement what you lack

• Build in public from day 1

Choose a familiar problem

The best products come from obsessing over a problem for YEARS, not weeks. So build what you've complained about for years

I've ranted about "businesses losing leads to slow responses" for 20 years and have been fixing it with various automation suites. TheSolarAI wasn't a "pivot." It was my years of obsession.

What problem have you experienced for 5+ years that's still unsolved?

And go Enterprise. Most indie hackers chase SMBs. It makes sense — I hate enterprise sales. Six to eight week sales cycles, security reviews, contract negotiations. I spent weeks in meetings instead of coding. But enterprise value is 10-50x higher.

Domain expertise compounds

And one more thing: If you're 35+ and think you're "too old" to be a founder — you're not.

I'm 49 and it's my unfair advantage: 20 years of domain expertise in a world of AI-first founders with zero real-world experience.

Most AI founders today:

• Graduated from Stanford/MIT in the last 5 years

• Know ML/AI deeply

• Have ZERO understanding of actual business problems

They "learn" an industry in 6 months by reading blog posts, interviewing 5 people, building a chatbot, then wondering why nobody buys.

I did the opposite.

Because domain expertise compounds; technical skills are a commodity.

If you're 35+, your 15-20 years of domain expertise is an unfair advantage younger founders don't have. Use it.

What's next?

My goals are centered on building a sustainable, high-impact technology platform that redefines efficiency in sales. In the short term, our focus is on securing more enterprise clients, scaling our leadership team, and expanding our multi-channel AI automation across both our specialized solar brand (TheSolarAI) and the new verticals under LeadTrackAI.

The medium term is about transitioning this into a robust, modular platform — introducing self-serve tiers, API access, and 10x-ing our infrastructure to handle massive volumes of leads as we grow.

Our long-term vision is where the two brands converge into a powerful mission: To establish the platform as the global standard for high-volume, automated sales conversion.

The specific goal remains to help 1 million households go solar by 2030. I'm building a profitable business that solves profound, real-world problems. I’m not optimizing for a valuation metric; I am building a company that champions impact, sustainability, and sanity as its core business model.

I post 2-3x/week about AI, enterprise software, solar, and indie hacking on LinkedIn. And check out LeadTrackAI!