Being a technology consultant and SaaS Founder, I get to speak with lot of people. I impart most valuable knowledge from them. I share those learnings here.
It’s not.
A developer friend recently started building an open-source ERP system out of curiosity.
A few weeks later, he told me:
“Every time I start something like this, I end up rebuilding the same things again and again… architecture, security, roles, permissions, background jobs, access control…”
Not because chat is bad. Because nobody reads old messages.
A student asks: “Can you explain this concept?”
Trainer answers. Two hours later, another student asks the same thing in a slightly different way.
Then another. Then another.
Let’s use open source like phpBB or MyBB to build a social network platform – That was the suggestion from a client
It sounds efficient. Existing software make launch faster and lowers the cost.
I asked him one question: “What percentage of your actual product will this really solve?”
Because this legacy software covered just 20% of what needed and rest 80% needs to be built on top of it.
A Chartered Accountant called me a few days ago.He was starting a CA exam coaching business and wanted a simple website.
Then he added, “Let’s integrate a payment gateway so students can pay online.”
Sounds logical, right? But here’s what I told him : “You’re new. No one is paying you online without trust.”
Early-stage businesses copy what funded startups do: Online payments, Automated flows, Full digital journeys. But that doesn’t work at that phase.
“More features = better product.” That assumption quietly kills most SaaS ideas.
Worked with a non-tech founder who had a long roadmap from day one : dashboards, reports, integrations.
But no users yet. He was asking me to build for scale… without knowing if the core problem even mattered.
I pushed back. I said “Let’s solve one problem. Properly.” Not 10 features. Just one that people actually use.
You’re not losing leads because of marketing. You’re losing them because you forgot when to reach out.
I spoke to a corporate gifting business. They had a potential to get leads automatically from their existing clients on same month every year
But they wait for clients to come back.Most don’t.Because no one follows up at the right time.
I suggested them an idea to build a CRM. It remembers events of the customer and auto generate leads 3 weeks before the event.
Both of them AI frameworks built in PHP. Both have everything — agents, RAG, chat, MCP. You can compare them to LangChain and LangGraph.
But one difference stood out fast.
Neuron AI → easy to follow
Symfony AI → harder to understand
Two weeks back, I was speaking with a client who wanted to build an Uber-like platform for logistics and transport.
He was deeply focused on the app flow.
How the booking screen should work. How tracking and UI experience should feel like.
But I told him something uncomfortable: What worked for Uber back then may not work for you today.
Adding a chatbot will make Product smarter. That’s what a prospect told me while discussing a checklist product.
His idea: Every item in the checklist should have a chatbot under it… so users can ask for clarification.
Sounds advanced. Sounds impressive. But there was a problem. This wasn’t an internal tool. It was meant for end customers.
These people want to complete a checklist faster and move on. So I said “don’t start with a chatbot.”
But , It usually isn’t a smart move.
Spoke to a team — 3 developers, 6 microservices.
Everything was “scalable”… except their speed.
So I told them: Microservices don’t solve scale of traffic.
