I am in no way a tech savy person, don't know coding, don't know networking or AI much either. But I definitely want to have a system like this. An AI powered gallery / video repository that can help me find moments, people, colors, objects from 100s of 1000s of files.
Local LLMs sound so cool but I know they won't be easy to setup or use for common joe like me.
Glad to be of help here. Let me answer the questions one by one.
- You're right about the confidence coming from investor expectations than actual enforcement experience. Startups avoid falling into patent litigation cycles as at early stages their primary goal is survival and gather funding. Health-tech requires heavy funding similar to industries like energy, or pharma. Having a patent secured innovation gives a positive signals to both potential partners and investors - that this startup has invented something new, and that their innovation is protected from being infringed. Whether a startup will actually enforce the patent in future highly depends on specific situation of their market. I have seen many startups (specifically the ones working on drug development platforms) more open to license their technology. Patents make that a little easy as it is a well known industry practice.
- It's mostly attorney driven as most founders who are working on their first or second ventures are not familiar with patent laws that well (unless they come from that background). This strategy is also only applicable in specific scenarios where it makes sense to keep the innovation hidden from competition (where you have lots of competing companies or where your work is catching lots of media attention).
- I haven't specifically seen this being discussed or got a chance to talk about it with a founder but I can share my experience here. AI/software patents are still surviving and only obvious patents are facing difficulties. You might have heard about Zoom vs Apollo dispute recently where court dismissed Apollo's motion against Zoom patents being invalid under Alice (My team is working on a story on that case separately). The patent has explained the inventiveness of the innovation being covered in a way that it is not abstract and actually showcase the improvement over prior art. If an innovation's inventiveness and improvement can be described in a non-obvious way (avoiding 'intelligent parser' and writing '0-5 point scoring system based on 5 parameters'), the patents can work.
Final thought about the last point: A patent's worth and cost is directly proportional to how important that innovation is for the industry (not for the startup patenting it). If a solution has multi-industry applications, it is not an incremental innovation but actually goes beyond SOTA, then a startup should opt for multi step protection, not just patent. Copyright the code, maintain trade secrets wherever possible, and patent the processes (if they solve the technical problems).
It highly depends on the industry but in health-tech, startups still opt for patent protection, simply because it works. For anything that is not suitable for patents, many founders opt for trade secret (as patents are open to everyone and can be used by competition to figure out how you're doing what you're doing).
Another strategy that I have seen experienced founder following is keeping their patent under examination for longer. They either respond late to the examiner's questions, purposefully opt for GAUs (Group art units. Different GAU reviews different tech patents) where the prosecution timeline is slower. This gives them extra time to keep their innovations hidden while still protected, until they are ready to enter market with viable products.
For AI based, software patents are a tricky game with many getting rejected under Alice 101 rejections. The drafting and prosecution methods also change a lot when dealing with software patents about AI.
I can share more in-detail if you have any specific question.
Context: I am not a founder but I work in a Patent and technology consultation company and we regularly help with prior art searching, patent monetization, and large scale patent analysis on industry.
I have talked to about 200 early stage founders last year through conferences or in-person interviews and talked specifically about this topic.
Would love to be looped in and can also work as a design partner if that is something you are looking for. I have been in this space for over 12 years now and have tested dozens of apps. Have worked briefly with Canva's team too back in the days.
Great app. It's at the perfect level of advanced and simplicity for the usecase you have mentioned.
I tried my hands on a few things and would like to ask / share feedback:
- Not sure if you're already on it or planning to, but target design/marketing agencies for it first. I know a few and they always have a chaos in their designing apps because they have to tell their team which brand they are designing a particular graphic for, and then find those specific templates, etc. Design process with your app starts from the brand which solves that problem.
- I currently work in a technology consulting firm and we regularly need to make multi-page reports, social media posts in carousal formats, or a series of graphics on the same topic. I noticed that currently your app doesn't support multi-page reports. I tested an A4 size, and did not see an option for social media post to generate more pages. That will be super helpful if you end up targeting individual brands for it later. Not having this could be a deal breaker.
- The design editor is really buggy at the moment. Totally understandable noting that you're at very early stages. But providing and option to "edit" AI designs and then a user really struggling to edit is counter productive. It did what MS word does to design when you try to resize and image and the element actually goes missing. I wasn't able to freely move elements on the page, or when I tried to resize an overlay element, it teleported to another dimension.
Overall, really great initiative and refreshing app. For generic usage like a single page social media post, a poster, or a featured image, this works amazingly well.
People generally don't spend time roaming around within the OS, or standing on desktop screen. I sometimes goes many days before I even lay eyes on my wallpaper. 90% of the time I open my laptop, browser is already open and I get to work.
Widgets are and always were a gimmick. User behavior won't change without a strong need. I don't think anybody need any widget. Nobody will miss them if they are gone.
I am currently running a website which is 100% written by AI, publishing about 2 articles every day. It ranks on Google for the keywords I am targeting and the traffic is also increasing (went from 0 to 2000 monthly visit in 6 months).
My content is not mindless however, it is highly specific to R&D folks, coming from annual reports, sustainability filings, and a company's public announcements. It is something not easily found outside but AI is helping me publish things on Scale.
The point is, AI or not AI doesn't really matter if your content fulfils a need, a purpose of your target audience. Google doesn't mind if it brings new value to the table.
You won't know until you start. It's not about writing book, it's just about putting your hardwork in words. You mentioned in your writeup that you used to think same way about building a console.
I have zero knowledge about electronics, consoles, or any of this and I am fascinated by what you did with your free time. Please write a book and publish it here. I'm sure lot and lots of people would love to read it.
Local LLMs sound so cool but I know they won't be easy to setup or use for common joe like me.