Real Insights from a Non-Creator Trying AI Video Apps
New tools, tricky trials, and emerging patterns in GenAI video
Recently, I was looking for AI tools to make videos for my personal projects. I approached this as a regular content consumer without any creator experience. Here are some observations and thoughts I gathered along the way. Please bear with me if anything is missing or unclear.
SEO still works for everyday users
For most users, discovering these tools remains somewhat limited. Exposure usually comes through social media posts with watermarks, friends talking, or online communities. I started with a Google search to find list pages that included various tools, since I would need to test them out myself anyway. OpenArt with strong SEO, and other apps with sponsored ads stood out, but it's worth browsing beyond the first few pages, as some early-stage startups may not be able to afford paid advertising.
AI-based search engines can help to a certain degree, but they still rely heavily on keywords rather than understanding actual functionality. YouTube recommendations can also be helpful because they often include demos that show how the tools work, the endorsements reflect the YouTubers' reputations, and since they're experienced with video tools, their opinions tend to be more credible. Ideally, there will be a low-cost and effective agent that can find tools, try them using the same prompt or instruction, and return a performance comparison.
Free trials for generative videos are rare
Compared to text and image generation, video generation is still in an early stage. Unfortunately, most platforms that offer these features don’t provide a free trial for general video creation, whether it is text-to-video or image-to-video. A few exceptions exist, such as Runway, Hailuo, Dreamina, and PixVerse. Kling and Pika, however, involved long wait times. Some tools offer limited styles, such as cartoon-only or avatar-based talking videos.
This leads to a familiar chicken-and-egg problem. Without being able to test quality firsthand, it becomes difficult to justify paying for the service. It will likely be resolved soon as costs decrease and performance improves. In the meantime, examples provided within the apps are quite useful by offering a quick glimpse of their capabilities. I have also started to wonder about feature usage and conversion rates of those apps, especially as A16Z data shows an increase in both apps and users in the AI video space.
New trends in conversion and consumption
Some of the payment models and conversion methods felt more like gamified personal consumption experiences, similar to entertainment apps rather than traditional productivity tools. Examples include getting discounts when canceling a subscription or earning credits by following social media accounts.
GenAI is making creative experiences more interactive and lowering the barrier to entry. This signals a future where the lines between creation and consumption blur. We’re seeing the rise of Prosumers: users who both produce and consume content. These were once two distinct groups, but they’re increasingly becoming one and the same. For instance, when a user dislikes a story and it is instantly recreated based on their input, they are simultaneously consuming and shaping the content. Sharing with social elements during the creation process can also spark inspiration and drive deeper engagement.
Test tools with clear goals
It’s easy to feel overwhelmed when trying multiple tools at once. Current video tools try to cover a wide range of needs, which can make it hard to focus. Keeping specific goals in mind helps with making trade-offs. Some features are impressive, but not always aligned with your actual needs.
For instance, Canvas streamlines the creation process and enables batch video generation by connecting to a database. A few apps offer avatar cloning, but the training results were not ideal and lacked flexibility for future updates. Using pre-made avatars within the apps remains the more reliable option. Leonardo’s image reference feature caught my attention, since image generation still lacks detail, consistency, and polish. These are just a few examples that came to mind and might be useful for your projects as well.
That’s it for this article and thanks for reading! I will continue to share more insights as I explore further.