Anything Goes : How a human and an AI partnered to build an end-to-end film suite. (Seeking feedback!) by Alexander Kruglov

As a producer, I wouldn't use such an app, and sorry to say that because I am sure you have put a lot into this. Largo.ai and others have done AI script analysis as part of their services for years, and currently a quick internet search reveals dozens of sites that offer AI script analysis. Everyone somehow assumes that a producer is interested in NOT reading a script and making their mind up, when their money or their employer's money is on the line. That's a strange assumption. You can usually tell if you are interested in a script in about the first 10 pages, and you know almost within the first two or three that you are NOT interested. After that, analysis from AI or anyone else is not relevant because the script is almost always going to be revised anyway in the normal course.

Shadow Dragu-Mihai I think you bring up some very good points.

Thank you for the comment. That is true, I wouldn't rely on this tool completely myself. This tool is more of a pre-production tool, for getting some ideas in different filmmaking areas, it is not for final production at all. I think more of an always accessible "professional" that you can consult, suitable for all filmmaking roles. I think of it more when the script passed initial stage and production begins. If the script is in use, then the tool could help, more as an personal helper. I would absolutely not rely on it as an marker that leads to an acceptance.

I would also like to add that the project is not a human replacement at all, it is a companion. And it is more aimed at mid-range and mostly indie filmmakers, with a lower budget.

In what way would be a professional helper?

I can not make a comprehensive show with screenshots of the app in the comments, but I think I can upload all of them to my profile. Briefly said the tool has three working categories for now. First one is Screenwriter Tools (Would maybe change storyboard to director). Storyboard with image generation, sfx, speech and camera angles proposals. Screenwriter's, Director's and producer's analysis, with selectable personas, suggests changes to screenplay and shows how much changes improves several metrics such as clarity, cost effectiveness, visuals etc. It also grades every scene from 0 to 100 with an overall comment and feedback. The final analysis runs all analyses together as a creative board, and selects choices that pragmatically balance out all the 8 metrics implemented and gives all rounder suggestions. Screenwriter can accept change by change, and the synthesize an pdf with one click, changing only the text they want. The file goes trough copyright screening right away to mitigate risks and.

The next section is Character Tools. It includes and comprehensive character analysis, with emotional arcs, how character drives the plot, and also description of character. Then it extracts metrics such as MBTI and IQ, and shows them next to score from 0 to 100. It also runs an personal tone analyser and suggests a dialogue refinements. When the analysis is run Actor Scouting is available. It is possible to select tier of actor, and how established actor is for each role, and the AI gives several options suitable for just that role, with percent suitability and casting director commentary

The next category is Producer Tools. First one is cost estimation. It is possible to select a budget tier for the project. The AI analyzes total budget (by minimum), gives ATL and BTL costs analysis, for every role and every production unit, scene by scene cost split, and comprehensive ROI analysis. The next tool is advanced risk mitigation. It presents risks with severity grades, and possible solutions, with impact on ROI (tries always to be positive, but some negatives can be for safety etc) and concrete suggestions on how to mitigate.

This is all for now. As you can see it is only advisory. All actions still must be executed by humans.

It is simplification of the workflow for one specific role.

Okay well IMO you are aiming a little high level for your actual market, and maybe including too much for that market. Perhaps cleaving off tools into different offerings? Here is my thinking, having 30+ years in the industry:

These might be of use to retail style social media content creators. Then again, that group generally drops the activity quickly and disappears or becomes sophisticated enough to not need the tools. Which is to say I don't know how you would market it to them or how sustainable that market is.

On the other hand, the services you describe seem pitched to a feature film team, which is actually something only an elite get to do, and a creative team at that level all run off to do their own functions - it's not centralized that way. There is no such thing as a one person feature film team. It takes several or dozens, so these functions are already in-built and separated. So the market for these tools which might be used among actual professionals is very small and entrepreneurial. Those people don't spend money on outside services which they can do and control in-house. There is the superfan group of amateur filmmakers but they don't have the money, nor generally the interest, to become professional and they don't appreciate the need to cover these functions. Essentially, the "content" creator group and the amateur film group and the professional group are different markets each with different understanding of the tools they need or use. That's a challenge for you in figuring out what tools are useful to what group and what group can and will pay for them.

So far as script analysis - the only people likely to care are aspiring writers who want validation on their work. Accomplished writers don't care about that, neither do accomplished producers. Coverage, etc. is marketed to the amateur writing community playing on their need for validation. When a legitimate producer requests it, it's within the context of large studios, where it's used as a filter. It's not of much use outside of that. So that's not, IMO, a useful tool. Also, by the time a producer decides to do a film, all the analysis etc. has been done and any more is irrelevant, as the script is going to go through revision after revision until, literally, post production is finished. So there's no daily purpose for this tool to a producer. Which is to say that as you describe the services, I don't see that they would do anything useful for professionals.

To be blatantly honest, one of the reasons Largo.ai and others have still been unsuccessful with selling these tools is that they themselves have no direct understanding of the market(s) they intend to service or the actual functions they are attempting to program AI to help with.

I don't mean to be impertinent or insulting, but have you ever produced a film from end to end? Even a short? If you have not become an expert yourself at those things, you have no frame of reference for what tool is useful or not, and you have no frame of reference to assess the quality of information your AI produces. In general, AI tends to produce unreliable results to a small or disastrous degree at least half the time, so if a producer is not an expert herself, it's problematic to use in an environment as expensive as film. Because that producer cannot spot the inevitable errors.

When you talk about ROI and "risk mitigation" especially, that's a planning and finance task, not a production function. The reliable training data for that for AI is specious at best, likely unavailable entirely. By its nature, those numbers are and always has been private - I am confident in saying that no reliable publicly-accessible figures exist, and private figures for one film are not applicable to another. Further, ROI and risk are affected by many non-quantifiable factors which in many industries don't exist but in film are implied and different in each project, and which are again, private. This means that the AI cannot train on actual data. That's not to say that the AI doesn't generate statistics. But it is to say that those statistics cannot possibly be reliable because there can be no reliable data for it to train upon.

Which brings me to another point on that - ROI and risk assessment are funny things. As a producer, I want to know the real on it for my own decision making. But I know that already, and I don't care what an AI says. The numbers, so far as needed, are dead simple to calculate. Almost faster than asking an app to do it. However, when I am raising money for a project, the ROI and risk assessment is what I say it is, and is legitimately different for every funding party I talk to because they all have different situations and concerns. Most of them don't care about the actual stats, to be honest. So I am not concerned with the predictions of an AI in that aspect and I don't see it as a tool I would rely upon except for being able to point to something that confirms what I am telling someone. That's just the way industry - all business in fact - works.