Flowcore ai
Automated customer service assistants
I got tired of watching small businesses drown in fragmented customer messages.
WhatsApp here. Email there. Webchat somewhere else. Nothing talks to each other.
So I built Flowter — one AI-powered inbox that unifies everything.
Stack:
- Next.js 15 + Supabase + Groq AI
- Multi-agent orchestration with smart handoffs
- WhatsApp (GoWA), webchat, email in one timeline
Hardest part so far: getting the AI to actually understand when to handle something vs when to hand off to a human. Two handoff max, then it's a person's problem.
What surprised me: SMBs don't want more tools. They want fewer. The inbox is the only interface they actually check.
Currently at private beta with real businesses using it daily. Revenue? Yes. Ramen-profitable? Getting there.
Biggest lesson: ship the simple version first. My v1 was just "inbox with AI replies" — no agent routing, no analytics, no knowledge base. That's what got the first paying customer.
What I'd do differently: start with WhatsApp only. Adding channels is easy. Getting the AI right is hard.
Any other founders building AI + customer service? What's your bottleneck?
2 Comments
Hello, and thank you for sharing Flowter.
I’ve gone through your post in detail, and the problem you’re solving resonates strongly with me. Fragmented SMB communication across WhatsApp, email, and webchat is something I’ve seen repeatedly in real systems, and unifying it into a single AI-assisted inbox is a very practical direction.
The part that stood out most is the AI decision boundary—when to act, when to escalate, and how to keep trust in the system. In my experience, that “handoff logic” is usually the hardest and most important part of these agent-based products, far more than the initial inbox aggregation itself.
I am a senior full-stack developer with experience in AI integration, backend systems, API design, and building production data pipelines. I’ve also worked on systems where real-time message handling and automation safety were critical concerns.
From what you described, I can see a clear path to improving and scaling the current architecture further, especially around:
agent orchestration reliability
escalation logic and guardrails
channel abstraction (starting narrow, as you mentioned with WhatsApp)
system observability for AI decisions
You already seem to have strong product intuition, especially around starting simple and focusing on WhatsApp first—that is exactly the right direction in my view.
If you are open to it, I would be interested in contributing as a technical co-founder or senior engineering partner, helping strengthen the system architecture and product scalability as you move beyond the initial beta stage.
Before anything formal, I would be happy to understand more about:
current system architecture and limitations
how AI routing is implemented today
what your biggest failure cases have been in beta so far
If there is alignment, I’d be glad to explore working together more deeply.
Looking forward to your thoughts.
One thing I'd be careful with:
The interesting question may not be whether SMBs want fewer tools.
It may be which problem they're actually hiring Flowter to remove.
Those sound similar, but they can lead to very different product and positioning decisions.
I'd be careful assuming the first paying customers are validating the same thing you think they're validating.