5 Months After Launch: What I Got Wrong About AI Video Tools

This is a really sharp observation. The gap between "I need to generate a video" and "I want to say something meaningful but I do not know how" is basically the difference between a tool company and a service company. Most AI products get stuck in the tool framing because that is what is easy to build and market, but the actual value is in the outcome.

I have seen a similar pattern building software tools -- users do not struggle with the interface, they struggle with the blank canvas problem. The moment you give them a starting point that feels personal, completion rates go way up. For something like emotional video messages, that probably means the first screen should not be "enter your text" but more like "who is this for and what is the occasion?" and then you generate 2-3 draft messages they can tweak. Removing the blank canvas entirely.

The "done-for-you emotional service" framing is interesting but I would push it further -- what if the positioning was less "we make videos" and more "we help you show up for the people who matter"? That reframe changes who your competitors are (not other AI video tools, but greeting cards and "I should call but I keep putting it off"). Completely different market with much less price sensitivity.

What does your current conversion funnel look like? Curious where exactly people drop off -- is it before they start creating, or after they see the output?

The hesitation before submitting is telling you something specific. People are afraid of getting it wrong when the message actually matters to them. That's not a UX problem, it's a trust problem. They need to believe the output will sound like them, not like a robot.
Done-for-you with a human review layer might close that gap faster than any feature improvement.

The positioning shift you described is the right one. "I want to express something meaningful but don't know how" is a much more honest problem statement than "I need to generate a video." The emotional barrier is the actual product.

Most AI tools underestimate how much the blank page problem matters. People aren't stuck because the interface is hard. They're stuck because they don't know what to say, and that's scarier than any UI friction.

The shift to done-for-you emotional service makes sense. The question is whether you can hold people's hand through the what-to-say part, not just the how-to-generate-it part. If you can make the starting prompt feel like a conversation rather than a form, I'd bet your conversion rate looks very different.

Still early is fine. The insight you just had is worth 5 months.

This really resonates. To answer your question — in my experience building an AI ad creative tool, users absolutely struggle more with knowing what to say than with the tool itself. We found that when we gave people a blank canvas and said "describe what you want," they froze. But when we flipped it to "just paste your URL and we'll handle everything," completion rates jumped dramatically.

Your insight about repositioning from "AI video tool" to "done-for-you emotional service" is exactly right. The best AI products right now are the ones that remove decisions, not add capabilities. Nobody wakes up wanting to use a tool — they wake up wanting the outcome the tool produces.

One thing worth experimenting with: pre-built templates or prompts that give people a starting point. Something like "celebrate a milestone" or "say thank you" as one-click options could help bridge that emotional hesitation gap. People often just need permission to start, not more features.

Very interesting problem you are facing. And actually, nice conclusion: people are using AI because they want AI to do the job for them.

I also have AI-ish product in video editing space, but my people are already coming with a ready video, so I'm this one step later 😅

This is a solid reframing. A lot of “low conversion” problems are really message mismatch problems: visitors understand the feature, but they don't feel the job-to-be-done strongly enough to act. When the trigger is emotional, reducing blank-page anxiety is usually more valuable than adding more AI output quality.

These kinds of reflections are super valuable.

Curious — was the biggest mismatch in expectations vs actual user behavior, or more in distribution and getting initial traction?

This reframe from "video tool" to "help me express something difficult" is a really important insight. I've seen the same pattern in other AI products — the ones that succeed long-term stop selling the technology and start selling the outcome the user actually cares about.

The hesitation you're describing reminds me of the blank page problem in writing tools. Notion, Google Docs, etc. all solved it with templates and prompts. For your use case, I wonder if showing completed examples first (like "here's what someone sent for a friend's birthday" or "here's how someone said thank you after a hard year") would reduce that emotional friction more than any UI change could.

One specific thing to test: instead of a single "create" button, try giving people 3-4 specific emotional starting points — "say thank you," "celebrate a milestone," "send encouragement," "express something difficult." When the user self-selects into an intent, the blank page anxiety drops significantly because they've already made the hardest decision.

What does your current onboarding flow look like? That first screen after someone lands is probably where most of the drop-off is happening.

This is a sharp diagnosis — you’re probably right that “what do I say?” is the true blocker, not video generation itself.

One practical test: replace “create a video” with 3 intent-first entry points (e.g. apology, gratitude, difficult update), then show a finished example before asking them to type anything. If starts go up but completions stay flat, the friction is still emotional safety, not UI.

If useful, I can run your page through a quick conversion roast focused only on emotional-friction moments (headline trust, first 15 seconds, and CTA anxiety): https://roastmysite.io/go.php?src=ih_softlywished_conversion_20260330_1951_hv

this is really well thought out. one thing i learned building outreach tools: the feature you think matters most is usually not the one users care about. have you gotten any feedback that made you rethink a core assumption?

the hardest part is always getting from zero to one. once you have that first user or that first dollar everything changes because you know the machine works. what would your first dollar mean for your roadmap?