Apex Utility AI

Community Showcases
21

@ fisj

Yes – the Utility AI concept was not invented by Apex, we are just making a good implementation of it for the indie game developers, using Unity. So, sticking to our core mission.

The Apex Utility AI is inspired by many sources. Dave Mark have made some really nice introductions to Utility AI (the example below is inspired by him), so we recommend checking him out for a simple explanation of the utility AI principles. We can also recommend checking out Guerrilla Games presentation of the Killzone tactical AI, which is a very nice example of utility-based tactical movement (we will include sample code in the Apex Utility AI for this type of reasoning).

Utility-based theory has been around for decades, and have been used to great effect in many economic theories for example. Utility-based systems has been used in everything from classifying mental diseases in psychiatry to locate where to dig for diamonds. It has been used to great effect in games such as Sims, Civilization, and the Killzone series. It is also closely linked to fuzzy logic, but perhaps more intuitively understood, since its based on scores rather than intersecting “zones”.

Fundamentally, it is a design-based AI. Meaning that you design the AI, as opposed to e.g. learning-based, search-based or evolutionary algorithms. Hence, it’s in the same class as finite state machines and behavior trees. However, the Utility AI can handle more complex decision-making, and can make decisions in areas were behavior trees and finite state machines just don’t cut it. For example is areas such as prioritization, spatial reasoning, incomplete information, or making choices based on multiple, dissimilar inputs.

The following is a classic decision-problems in e.g. an FPS game, and is often used to explain how Utility AIs work:

Deciding on the right action, can be based on scoring each action. However, the problems is that basis for decision-making can be very different for each action. Taking cover can e.g. be depending on the health of the AI and the distance to nearest cover. Firing can be based on the proximity and number of enemies, or e.g. good firing opportunities. And whether the load the weapon be based on the current ammunition count in the weapon etc. All very different inputs that need to be compared.


The Utility AI solves such as problem very elegantly, either with curves (such as the above) or with point scoring systems (see e.g. below). The latter is sometimes easier to understand, design and implement, so we end up using the paradigm a lot in the implementations we have done in the process of making the AI. We will have several examples with the Apex Utility AI, where we apply these principles to many complex problems of AI. Earlier in the thread you can see some of the areas we will publish tutorials and concrete AI implementations for in the coming weeks / months.

The above example is very simple, in many of the games we have working on, you need more complex decision-trees and decision-matrices to make really good AI. Hence, the Apex Utility AI comes with several tools to making complex and good AI simple. Including advanced editing, parallel execution, out-of-the-box behaviors, best-practices, visual debugging etc.

You can example of an AI here:

Interestingly, we also found in our research prior to making the utility AI product that the games with the best reviews for AI, and at the same time the games where the AI programmers were most satisfied with their AI, happened to be based on Utility AI systems.

So, in many ways utility AI is a hidden gem, which is why we wanted to bring it to the indie developers.

So stay tuned, and let us know if there are concrete problems you are dealing with in your game, that you want the AI to solve. It’s our hope the Utility AI will enable a whole new generation of games with great AI.

/The Apex Team

2 Likes
22

The first tutorial video is uploaded “Survival Shooter Introduction”
https://www.youtube.com/watch?v=WAIbq0CmuZA

I have created a Survival Shooter playlist, we will be adding videos over the next coming days as soon as they are ready!

Remember to subscribe!

Let us know what you think!

/ The Apex Team

23

The 2nd Video is up!

/ The Apex Team

24

I read your article on the website

You say Add-on FPS, RPG, RTS…

I would like to know in detail

When we can see it?

25

Hi Cartoon Mania,

It’s submitted to the vetting team, so it should be available anytime soon.

The core package is released with several AI examples. There is a complete survival shooter tutorial that covers many aspects for third-person and first-person shooters such as target acquisition, tactical movement and action prioritization. Over the next weeks additional tutorials are released.

We don’t have the release dates for the full add-on packages yet, but we are currently working on an RTS demo project.

/The Apex Team

1 Like
26

thank you
I am very interested in the auto battle
This is a good solution for automatic battle?

I am a little worried about your business model…
Because it is a new way

I look forward to full package add-ons.
but I do not want to wait for long

27

mm yeah, was quite intrigued about this but that business model is a deal breaker for me.
I would’ve understood a higher price, but subscription no thanks, got too many of those already.

28

Agreed. Software costs for small game developers are already out of control. No way anyone without big funding can afford multiple indefinite subscriptions for a team. We love and use Apex’s other products on a regular basis, but I can’t support this business model as it leaves small teams like ourselves out in the cold. Yes we could afford this one single subscription, but certainly not if all the assets we use are subscription based.

@Jakob-Rasmussen , another major concern with the subscription model: you are holding your customer’s projects hostage. If for any reason the subscription falls through (can I trust Apex to not have billing or login issues - ever?), I assume we’ll lose access to these tools and the ability to continue development. This is unacceptable.

If you absolutely refuse to give up the subscription model, please at least take a cue from Unity and offer the full product for a one-time price in addition. As it is, this asset will not even be allowed on the asset store, since Unity doesn’t allow this business model. For any potential devs looking at this product: ponder this fact before signing up.

I am very disappointed that Apex has chosen this model, as I was looking forward to this product and would have paid 10x the subscription cost for a one-time purchase. As a fellow asset store publisher, I understand the desire to avoid Unity’s 30% fee and bring in more consistent cash flow, but this is not the way to do it. Instead of purchasing this asset, we will be using Behavior Designer.

2 Likes
29

Hi @donov , @vvander_1

Thanks you for the feedback on the subscription model – very much appreciated. We are very interested in this dialogue, so we are looking forward to comments and feedback on how we can find the best model for all parties.

Below I will explain the reasons for launching with this new model.

Apex Utility AI is a professional AI product based on many years’ research, prototypes, and experience from concrete game development projects. It is a professional product with optimized performance that allows even the tiniest indie developer to ship pro-grade AI. It comes packed with industry-based best-practices, tutorials, code examples, and a full tutorial project. Furthermore, Apex offer top-of-line support for no additional costs. That is the Apex mission: We want to ensure that indie developers have pro-grade software at their disposal to make great games.

That also means, Apex is not meant to be cost-leaders, or the lower-priced product on the asset store. There are a lot of great assets that sell for very little. Apex is meant to be an investment that in the end allows you to reduce the cost and time in developing games, and ensure that you in the end can ship a game with much better AI. I.e. if you choose Apex, you are guaranteed a great asset, great support and - in the end - a greater game at a lower overall cost.

Being able to do that comes at a considerable cost in development and operations. To ensure that we can still provide this quality of asset at prices that are affordable to any indie developer from tiny to large teams, we need a model that distributes the cost fairly across the developers. The one-man team with a small budget should pay very little, the big well-funded team should share a bigger part of the burden.

We absolutely understand your concern with the subscription, and we have received other requests for the option to have a one-time fee – very similar to Unity’s offer to pay per month or pay a one-time fee for a version. We also understand that a new model can lead to scepticism until it has proved its worth.

The concern about holding projects hostage is also a valid point. There are many software companies that suddenly forces you to upgrade for a fee, or ship a new version that renders all your add-ons or downloadable content obsolete.

The challenge here is to find models that ensure indie game projects the best assets at a price they can afford, while still ensuring that the asset developers have enough cash flow to continue developing great assets.

On this basis, we analysed the indie projects’ cash, workflow and purchasing patterns and compared this with the flows of bigger projects. We also spoke with many of our customers about their requirements.

These are some of the challenges that we are seeking to address with this new model:

#1

Indies often need one license, whereas professional projects need many. However, the current model on the Asset Store - anything but Editor Extensions - is a full site license. The consequence is that indie projects pay a very large share of their budget for the same asset that a professional studio use for 40 developers. We think the per-seat subscription is more fair to indie developers, as the big studios can bear a larger share of the develop costs.

#2

Indies often have shorter projects, which can be down to anything from 1 to 3 months for the AI part. With the subscription model you only pay for the months you need. Hence, if a project has 3 man-months of AI development, you only need to pay for these months. That frees up a lot of budget. However, on a big, well-funded project, you might need more man-months for AI work, and hence you are also able to pay a large part of the share of development costs. Again, we think this model is fair to indie developers.

#3

The Apex AIs still run after you stop paying for the license, as license is tied to the editor. We have of course designed our products to facilitate the best experience for the developers and not disrupt their workflow. Hence, your project will of course still run if the Apex is unavailable. Your game and AI will still run when you stop paying the license.

#4

The subscription model has the advantage that we can ship upgrades etc. automatically, without you having to buy them separately. Our experience is that it provides a better experience for you. You can then add additional seats / licenses when the develop team scales up, and still be able to predict costs etc. Some add-ons to e.g. visual studio become second nature to keep subscribing to, because they are very good and saves you a lot of time. This was always the vision the Apex: We want to save time and cost for indie developers, so they can ship better games, faster and with lower budgets.

#5

With the licensing model, we can better avoid piracy. In a way we felt oddly proud that someone wants to share Apex product on torrents. However, the downside - of course - is that the paying indie developers have to share a bigger part of the development costs. By avoiding piracy, we can overall lower prices as well as continue developing great software without going bust.

#6

With the licensing model we get the ability to ship trials. With a subscription model we can offer a trial and that means you can work with the full product for a full month to see if it delivers as promised, before you have to pay anything. And, you can cancel the subscription and not pay anything if the product does not fulfil its promises. Hence, there are no caveats with this model and it’s a great service.

In conclusion, if we want more great assets on the asset store, we need to find a model that can sustain asset developers, without breaking the back of indie projects. We at Apex are willing to risk our neck to get a better model for the indie developers and the asset developers alike. We thought the subscription model offered a number of advantages to distribute costs more fairly. If this is not the case, we would really appreciate input on what other models could be preferred, that solve the problem in a better way. But we also really hope you – the indie developers – will support us in this mission.

If the solution really is a one-time fee that could also be an option. But, if you look at Unity’s pricing the one-time fee is x20 the subscription price for a per-seat license for two devices, and that is without the service and support you get on an Apex asset on the Asset Store. That could end up being quite tasking for an indie development budget, as the money has to be paid up-front, whereas the subscription model can be distributed over the life-time of the project.

Hence, we really hope you download the trial and give the subscription model a try, and we really look forward to comments and input on how we can improve this.

Best Regards,

Jakob

1 Like
30

I know APEX UTILITY AI is Excellent and awesome…
I was really looking forward to this product
I think that monthly fee is even inexpensive

but for me there is principle

I buy only bestselling products.
best selling product generate revenue
So supporting will continue.
if there is no revenue Supporting is broken

I hope you sell a lot of products.
You are good developers but You are not good marketers.

First, you must be a bestseller. then You can do new experiments.
You must create an enthusiastic followers. And then try a new business model
Be a star of asset store then you can do what you want

Please think of your brand value…

I think APEX is not enough to try something new

Right now I do not need this product.(I wait Add-on KIT) but I want to help this product.
If sell for 100 dollars right now, I’ll buy the product.

If you insist on a subscription model… I must wait when the product is a bestseller.

Unfortunately, It will not be easy to be bestseller

except for The disadvantage of the subscription model. You are do not have enough marketing skills to sell the product.

31

Looks like a really interesting AI package, unfortunately, I’m not much for subscription models as in a crisis situation one may not have the funds, it’s why I paid for Unity Pro in full, and will only use a standalone version of photoshop, and will buy Amplify texture 2 rather than subscibing to granite. Pity you can’t give the option, I know quite a few would prefer to pay monthly, however, there are people like me that have been in situations several times that paying a subscription model would just mean a game could not be finished or the AI would have to be swapped out again. If it proves to be the bees knees of AI I might reconsider but at the moment I will stick with Behaviour Designer.

32

Hi @Cartoon-Mania

Thanks for the valuable feedback.

And thank you very much for the precise quote on what you consider a good price on the product. This is critical information.

We will definitely continue support for our products, as we in this for the long run. But we do take your feedback very seriously, and we are considering whether we should offer a buy-for-good option just like Unity does.

We also appreciate your points on how we market this, and we are taking note of this both new releases to the Utility AI and for future products.

Under all circumstances, we still propose you consider downloading the trial, and we are very interested in concrete feedback on the functionality you are looking for.

Best Regards,

Jakob

33

Hi @sicga123_1

We appreciate the feedback. It sounds you have been through a very nasty process with previous project, where you had to swap out the AI. If you are willing to share the story, it could be valuable to other developers as well.

As I wrote above in reply to the other posts, we are considering a buy-for-good option.

I want to iterate, that the Apex Utility AI keeps working after you stop subscribing to it, you just cannot use the editor. Once you subscribe to it again, you can edit your AIs again. Of course that does not help you if the company behind the product is no longer there to support the product or if the licensing server is taken offline. However, we don’t foresee that happening to Apex, as we are a solid company.

Did you download the trial? I think it would valuable to have your feedback on the actual Utility AI product.

Best Regards,

Jakob

34

@Jakob-Rasmussen

As someone already said before,
I like to buy best selling assets and assets that show potential for future development.
That’s why I very often take the plunge in to beta or early versions to help a developer that has a promising asset that I know will deliver(Apex Path being one of them and waiting for a justifying reason to get the add ons as well),
and also take advantage of the lower introduction price while in development.
I would be ok paying even full one time higher price if the asset is worth it, $100 for example.

To be honest, the current business model introduced, puts me at unease even for continuing using ApexPath, as it’s very much possible you will introduce the same model for it as well in a new version

Regards
Igor

35

@Jakob-Rasmussen Thanks for the reply. You have some good points, but ultimately the most important point is the one that you have reiterated a few times above:

This wasn’t mentioned in any of the marketing on your website or this thread - I recommend you add this as a major selling point! Subscription-based assets aren’t a wholly unique model… Speedtree is another example, and on their website they state specifically that only the creation/editing capabilities are disabled when a subscription lapses.

However, many developers like ourselves will refuse the subscription model for software outright, so you’ll probably still lose some customers if you can’t offer a non-subscription version. Like I mentioned above, we are willing to pay many times the subscription amount to have guaranteed access. Hopefully Apex will consider this, as we love your other products and would almost certainly purchase this asset outright if possible.

P.S. I’ll do you the favor of not calling you out more than this on the piracy excuse: This isn’t 2010! Pirated downloads aren’t lost sales, and all devs should know this by now.

36

Thank you very much for your support on the discussion of this model.

We are still considering the best way here. We actually saw the subscription model as an advantage for the indie developers, and we had so far not received any negative feedback on it from our beta users. Hence, we want to be transparent about this all our users.

Hence, with the recent viewpoints, we are very interested in some feedback on this point.

Some questions we have are:

1/

For a one-time fee, what would be a decent price point for you? Some things to consider here are, whether this fee should include add-ons, support, and other items. And also, how often would it be decent to launch new version, for e.g. a 50% upgrade price. If you look at Unity for example, you pay a new fee for every major version, and buying a Unity license is equal to 20 months of subscription.

2/

Another option is a subscribe-to-buy model. I.e. here you subscribe, and when you have paid a fee equal to e.g. 20 or 24 months, the software is yours to keep. Would that be an interesting model?

3/

Another thing we have considered is: Currently, we offer a free trial for one month. We don’t see other packages doing this. I.e. you can actually try the Apex utility for a full month with full functionality at no cost, and you can cancel at any time with no cost. Is this something that you see as being useful?

4/

We do require a credit card to verify identity when starting the trial. Similar to the model used by e.g. Netflix. Is this off-putting or does it pose a problem when considering the Apex Utility AI?

Best Regards,

Jakob

37

Personally, I’d say $100-125 is a good price point from what I can see of this asset’s features via marketing material. Even this might be a little high though… Again this is based on the marketing alone, not the trial, though. I could easily see the price of this package being higher depending on the feature set, but considering how powerful UFPS is for only $75, and your biggest competition on the asset store (Behavior Designer) is only $75 as well, going higher will be tough. Playmaker is the best-selling asset and is extremely complex, but even they can’t get away with charging over $100 for that asset though. All of these packages provide free updates, too. Charging for new versions will be difficult unless they are major overhauls - and discounts for current customers will also be necessary.

I realize that the prices I quoted may seem undervalued to you, but I have seen firsthand the results of pricing our assets lower than we originally felt was fair: after decreasing prices to 1/3rd the original amount, we had 5x the sales! This kind of thing is seen on Steam, too. Gabe Newell is famously quoted as saying that discounts produce more revenue in a digital marketplace.

I’m not sure comparing your asset to Unity is a good idea. Unity is a full engine, ecosytem, etc, and is therefore difficult to replicate. They are really only competing with Unreal for their main market segment (for the most part). The demand for their product is inelastic, while the demand for your product is very elastic at the moment - basic economics dictates that Apex will need to have lower relative prices to remain competitive. Basically, you can’t charge 20x sub price for complete access because people will just purchase something else instead. There’s less of a choice with Unity, so they take advantage of that. If you were interested in the subscription model, I think it would have been better served to introduce something like this AFTER finding the best price for purchasing it outright on the asset store. If you need to compare yourself to Unity, remember that they didn’t offer the subscription model until many years after the original launch. Their sub price was based on the full price, not the other way around.

Definitely something to be considered. Especially if you’re unwilling to stay under $100 with the outright purchasing price. Sub-to-buy is still a subscription, though, and any devs that have the mindset of our studio will not go for any subscription. We only purchase software outright.

Definitely useful! I think this is a fantastic idea, in fact, since for editor extensions you can’t offer much of a demo. We would have already installed the trial and been using it in our projects if the subscription model wasn’t the only purchase option.

I don’t think it’s off-putting, personally. It may be to others, however - I can’t really speak for anyone other than our studio, though.

P.S. That’s all the free business strategy consultation I can offer :wink: I have to get back to doing economic evaluations of our company for now. Will definitely be checking this thread for updates however - good luck and hopefully we will be able to use the Utility AI package at some point.

1 Like
38

People are not too interested in this product. So there are no feedback(negative and positive)

Do you have the desire to sell the product?

It is responsible for all of you.

Please Show us Cool demo

your shooter tutorial is not impressive.

There are many products like that on the market.

Do you plan to sell APEX AI in Asset Store?

39

@Cartoon-Mania

Thank you for the feedback.

The Survival Shooter is meant as a tutorial to outline the principles in the Utility AI. It comes with complete and detailed documentation and is aimed at beginners.

Over the coming months we are releasing a number of additional tutorials showcasing what you can do with the Apex Utility AI. We are also - in due course - releasing a number of game frameworks that are aimed at vastly reducing the time for developers to make great games.

Beyond this, we are very interested in understanding whether you have concrete projects you are working on or concrete AI behaviors that you would like to see? Perhaps you have something that you have been struggling with that we could use the Apex Utility AI to solve? Your feedback is highly appreciated.

Yes - the Apex Utility AI will arrive at the Asset Store as soon as it goes through the vetting process.

Looking forward to hearing from you.

Best Regards,

Jakob

40

Jakob Rasmussen

Thank you… I madly want to use this product. but I think Unpopular product loses supporting.

I worry about it.

I want two things.

  1. Real World AI
    I know that you are ready for real-world examples.
    Is this Sims- Like -AI Right?
    In fact I am selling of assets.

Girl Character Customization Set
Boy Character Customization Set
Cartoon Furniture and Interior Set

I want to make simple games like Sims using my assets
of course Sims is a very complex game. I say a very simple game like Sims.

  1. AI VS AI Auto Battle

I’d like to see a thrilling battle AI VS AI

I think Auto-Battle is the most difficult and beautiful AI

I know the mandate . The battle was very interesting.