I built Formaly because I kept getting annoyed by forms.
Not the idea of forms, forms are useful. The annoying part is how much important feedback gets trapped inside bad form experiences.
You send a survey because you want to understand users. The respondent opens it, sees a wall of fields, and rushes, skips, abandons, or gives one-word answers. Then you export the data and spend hours hunting for the actual signal. Both ends of that are broken, and they are broken in ways the data confirms: drop-off rises sharply once a survey passes ~12 questions or 5 minutes, and most open-text "analysis" is really just someone scrolling a spreadsheet hoping a pattern jumps out.
That felt fixable.
The original idea
The first version was simple:
What if creating a survey felt like explaining what you want to learn, and answering it felt closer to a conversation?
That became the core of Formaly. You describe the goal. Formaly generates a structured survey. Respondents answer through a chat-like interface or a traditional form view. After responses arrive, AI helps summarize the themes and surface what matters.
Why conversational forms
People do not naturally communicate in dropdowns and radio buttons. They explain, they qualify, they tell stories, they say "it depends."
Traditional forms are good at structured data but weak at nuance. Conversational surveys can ask one question at a time, adapt to answers, and feel less like paperwork. And this is not just aesthetic preference, research on AI-moderated conversational surveys finds they produce more thoughtful, more actionable answers than static forms, with most participants preferring the experience.
That does not mean every form should be a chat. It means feedback collection should have more than one mode.
What Formaly does today
Formaly lets you:
- Generate surveys from prompts, build them manually, or import from a PDF
- Use 30+ question types, from NPS, matrices, and ranking to image choice, sliders, and constant-sum
- Run surveys in chat mode or form mode
- Embed surveys on websites or trigger them inside your product with the SDK
- Send tracked, personalized email invites
- Analyze responses with AI summaries, sentiment, per-question breakdowns, completion funnels, maps, cross-tabs, and quality checks
The product is still young, but the direction is clear: help teams ask better questions and understand the answers faster.
Under the hood
The current stack:
- Next.js and React (App Router, React 19)
- TypeScript end to end
- PostgreSQL with Drizzle ORM
- Better Auth for authentication
- Tailwind and Radix for the UI
- OpenAI-compatible model infrastructure through Nebius
- Redis for rate limiting and caching
I care about the foundation because survey tools handle real customer data. The product has to be fast, reliable, and boring in exactly the right places.
The practical lesson
I do not want Formaly to be "Google Forms with AI sprinkled on top."
I want it to be a feedback system: creation, collection, analysis, and follow-up working together. Forms should not be where good feedback goes to die. They should be the start of a useful conversation.