← Back to home
Product Case Study • Market Research • Pivot
Flowly
From IT helpdesk bot to AI-powered customer support for solo founders. A study in finding real problems worth solving.
Role
Solo Founder
Timeline
2024 — Present
Stack
Next.js, Neon, Cloudflare, Slack API
Type
SaaS • Open Source
01
The Original Idea

An AI-powered IT helpdesk bot for Slack, targeting small companies with 50–500 employees.

The thesis was simple: these companies don't have real IT support systems. Their one IT person is drowning in repetitive questions — password resets, VPN issues, "is Zoom down?" — and enterprise tools like Jira Service Management and ServiceNow are too expensive, too complex, and too heavy for a team this size.

The product would sit inside Slack, intercept IT questions, answer the easy ones automatically, and only escalate to the IT person when it couldn't help. No portal. No ticket forms. No behavior change. Just support where people already are.

Technical Architecture
Cloudflare Workers Gemini / GPT Event Sourcing Next.js Admin WebSockets Okta / Azure AD
Pricing Model

$300 base + $2/user/month, with a $500 monthly floor. Targeting ~$67K ARR in Year 1 with 8 clients.

The architecture was solid. The market thesis wasn't.
02
Where It Fell Apart
01
The Market Was Already Crowded — Or Empty for the Wrong Reasons
The IT helpdesk space has well-funded incumbents: Jira Service Management (free for up to 3 agents), Freshdesk, Zendesk, and Moveworks ($305M in funding). For the 50–500 employee segment, the real competition wasn't these enterprise players — it was doing nothing. Most companies this size handle IT through Slack DMs and shoulder taps. If they've never prioritized it, selling them a tool is an uphill battle.
02
The Unit Economics Were Razor Thin
At $500/month per client with estimated infrastructure costs of $275–475/month across 8 clients (AI API calls, Neon Postgres, Cloudflare Workers, Trigger.dev), margins were nearly nonexistent. The business needed significantly more clients to be sustainable, but the sales cycle and onboarding burden made rapid scaling unrealistic as a solo founder.
03
Niche Exploration Revealed Entrenched Players or No Budget
K-12 education had Incident IQ dominating 15,000+ schools. Nonprofits had free tiers from Freshdesk making price competition impossible. Healthcare required HIPAA compliance unviable solo. Law firms were owned by MSPs like All Covered with 1,000+ relationships. Construction was underserved but hard to reach and slow to adopt.
04
User Research Pointed Somewhere Else Entirely
A conversation with a DevOps engineer at Blue Origin revealed that his primary pain wasn't ticket management — it was incident response and being the single point of contact when things go wrong. The real pain isn't about tickets. It's about being the bottleneck.

That realization shifted the entire product direction.

The pivot.
03
Customer Support for Solo Founders

The vibecoding movement created an entirely new persona with an acute, unsolved problem.

Thousands of non-technical or semi-technical founders are now shipping SaaS products using AI coding tools. They have paying customers, but their support process is chaotic — a personal email inbox, a Discord server, maybe a shared Slack channel. When customers have questions, the founder answers them manually, one by one, while also trying to build the product.

Existing solutions don't serve this persona. Intercom, Zendesk, Freshdesk are built for companies with support teams — overkill and overpriced for a solo founder with 50 customers. Chatbase, DocsBot, and SiteGPT are simpler, but require manual knowledge base management that solo founders will never maintain.

Solo founders will never maintain a knowledge base. The product must maintain itself.

The Proposed Product
  • Onboard via GitHub, not document uploads. Connect your repo, and the bot reads your codebase, README, and commit history. As you ship updates, it stays current automatically.
  • Learn from founder corrections. When the bot gets something wrong, the founder responds manually. The system observes and incorporates it — next time, it handles it automatically.
  • Zero-infrastructure support. No ticket system. No agent dashboard. A script tag on your site, a Slack/Discord integration, and a simple dashboard. Setup in under 5 minutes.
  • Priced for indie scale. $29–49/month — trivial for a founder making $2–5K MRR, far below Intercom's pricing.
04
Key Decisions
Abandoned enterprise IT helpdesk positioning
Market either had entrenched competitors or no willingness to pay.
Rejected DevOps / incident response pivot
Trust barrier too high for AI executing infrastructure changes; wrong market for a solo founder.
Chose customer support over internal IT
Larger market, more accessible buyers, clearer willingness to pay.
Targeted vibecoders over established companies
Underserved persona with acute pain, reachable through community rather than cold outreach.
GitHub-based onboarding over document upload
Core differentiator against Chatbase; eliminates the maintenance burden that kills bot accuracy.
Learning from corrections over manual retraining
Solo founders won't maintain a knowledge base, so the product must maintain itself.
05
What I Learned
01
Architecture without validated demand is expensive procrastination.
I built event sourcing, adapter patterns, and encryption schemes before talking to a single potential customer. Every hour on infrastructure was an hour not spent on discovery.
02
Your first market thesis is almost always wrong.
The original target sounded logical but fell apart under scrutiny. The real opportunity came from following the research, not from the initial assumption.
03
The best product advantage at small scale is giving a shit.
Jira and Chatbase serve millions. They can't offer white-glove onboarding, same-day bug fixes, or custom integrations. A solo founder can, and that's a real moat until you outgrow it.
04
Talk to users before you write a single line of code.
One 10-minute conversation with a DevOps engineer invalidated an entire product direction. Five more with the right persona could have saved months.