Utility AI - Smart NPCs, Enemies & Living Worlds (Wise Feline Ultimate)
NoOp Army
$165.00
Jump AssetStore
Ship enemies and NPCs that decide for themselves. Utility AI with influence maps, smart objects, and memory & emotions: believable, non-scripted behavior without manual creation a behavior trees.The package itself is compatible with all render pipelines but the materials of the demo scene are not, which doesn't matter for your projects.Since the package uses UI toolkit for the editor window, You cannot backport it to versions before 2019 which don't support UI Toolkit. We did not test in versions lower than 2020.3.Our new demos can be converted to URP and HDRP as well.★★★★★ "The best utility AI system on the asset store."★★★★★ "Will save you weeks of rolling your own utility AI system."★★★★★ "Tried building my own, now back to WiseFeline."Ship NPCs that decide for themselves: enemies that pick their own tactics, villagers that live their own lives, companions that react to what is actually happening. No hand-wiring a behavior tree or state machine for every case.Wise Feline is a complete AI suite built on utility AI: every action gets a score, and your agents always pick the best thing to do right now. Add influence maps for tactical positioning, smart objects for interactive worlds, and memory & emotions for characters your players will remember.Build your first AI without writing code. The ten-minute combat quickstart builds a soldier that patrols, spots you the moment it has line of sight, chases you, opens fire, retreats when it is hurt, and heals at its camp before walking back out. Every piece comes from the built-in actions and considerations (with animation integration); there is no custom AI code in it at all. Our full life simulation and ecosystem demo needed fewer than 10 custom classes. When you want full control, everything is clean, documented C#.Try the Lite version for freeEverything included: over $140 of modules in one packageInfluence Maps ($60): tactical and spatial reasoningSmart Objects ($45): intelligent, moddable world objectsRemembrance ($35): memory & emotions for NPCsBlackboards and Den (visual debugger & data recorder): included and integratedUltimate-only features on top of the free version: multiple selection mechanisms, priorities, and score deltasWhat you can buildEnemies and squads that weigh every option, every tick: never stuck, never out of good choicesNPCs with their own goals, needs, memories, and moods, living their lives whether or not the player is watchingAgents that react to unexpected situations in non-scripted ways that surprise even their designersAI you can redesign mid-production: add and remove actions freely, with no fragile trees or transition webs to untangleBehavior that can be modified procedurally at runtime, by your systems or by your players as user-generated contentStrict sequential actions where you need them; emergent behavior everywhere elseTactical and spatial reasoning with influence mapsAnswer questions like:Where should I stand to stay close to allies but away from enemies?Where should I throw the grenade to hit the most enemies?Which routes do cargo caravans travel, so the bandits know where to strike?Where should procedural generation place the next house, given population and building density?You can also use influence maps to record where events happened, or to drive procedural content placement.Smart objects: put intelligence in the world, not just the agentObjects tell agents how to interact with them: agents request what to do, objects decide howDesigners add new objects without ever touching character codePlayers can mod your game with new objects that just work with existing NPCsObject behaviors combine in ways even the designers didn't anticipateWhy utility AI, and how this is different from a "utility node" in a behavior treeBehavior trees and state machines are the right tool when behavior must be strictly scripted. But when agents should feel alive, utility AI is how you get there; it is the approach used in The Sims, Guild Wars 2, and many other major games. A utility selector node inside a behavior tree only scores a handful of branches. Wise Feline is a complete utility AI authoring model: considerations with response curves, per-target scoring, priorities and selection algorithms, plus a knowledge stack (AI tags, influence maps, smart objects, memory) that gives agents real awareness of the world. Your designers tune curves and importance instead of rewiring graphs, and the visual debugger shows exactly why an agent chose what it chose.What can Wise Feline do for your genre? Click here.High performance. Burst-powered influence maps (optional). Suited to open-world games, simulations, emergent gameplay, virtual worlds, and MMOs. Flexible enough to drive cameras, lighting, and VFX, not just characters.Discord | Documentation | Forum Thread | X/Twitter | Video tutorials | Download the playable living-world demoSubscribe to our newsletter for news and sale announcementsBuilt-in actions and considerations for fast prototyping: A growing library of built-in actions and considerations (movement, patrolling, targeting, spawning, animation integration and more), generic enough for prototypes and for real games. The combat quickstart in the manual builds a complete enemy loop (patrol, chase, attack, retreat, heal) from them without writing any AI code.Two sample scenes: A ten-minute combat quickstart with two soldiers built entirely from the built-in library, and a life simulation sample with cartoonish artwork which shows how the systems work together in a larger scene.Easy to learn: There are only a few concepts to learn, mainly actions, considerations and agents. Each agent has a set of actions to choose from, and each action is scored by scoring a set of considerations which show how important the action is right now.Easy programming interface: Derive from ActionBase and ConsiderationBase to create new actions and considerations, and drive your agent from your action code. A single Brain component on the agent runs the whole thing.Visual debugger and design tool: A window for designing your AI by setting consideration curves and choosing how much each consideration affects an action, and for debugging action scores at runtime. You can examine the exact score of every action at every frame and see how to change the curves to get the result you want. Your agent doesn't set enemies on fire as often as you would like? Make the curve of the Love Of Fire consideration steeper.Action priorities and interruptibility: Each action can optionally have a priority and can declare itself uninterruptible. Actions with higher priorities are preferred to actions with lower priorities regardless of score. Priorities are available in the ultimate version.Multiple action selection modes: Each Brain can pick its action with different algorithms: highest score, weighted random, or random from the top N, each with or without priorities. Selection modes other than highest score are available in the ultimate version.Action based targeting: Each action which needs a target is scored once per target and the best target is chosen. Finding targets is just a callback: use a manager class which knows all the targets, a sphere cast, the AI Tags module, or any system of your own. A per-action upper limit on target count keeps the system from spending too much time looking at too many objects.Influence maps: Spatial queries like: where are there no enemies but plenty of trees to cut? Where should I throw the grenade to do the most damage? Which paths do cargo caravans use, so I can set an ambush? Includes optional decay, line of sight, and Burst support.Blackboards: Share information between different parts of your AI and build reusable actions and considerations. Included in both versions.Data recorder (Den): Record data per GameObject per frame as a tree of rich text and inspect it later. It integrates with any script which wants to know what happened over time, and it is already integrated with the brain, so all action and consideration scores are recorded for you.Remembrance, memory and emotions: NPCs remember what happened in the past and hold emotions about characters and objects. Want an NPC to rage-attack at low health because you killed their pet? Want the whole village to remember it and raid the player? This module is how.Smart objects: Put the behavior in the object. The agent only asks which objects can fulfill a need, the objects advertise themselves, and the agent uses one of them to perform the task.AI Tags system: Tag objects and query for objects in a sphere fast, using an octree instead of raycasts. Want all food around an NPC so they can eat? Search for objects tagged food around their position. Add tags like enemy, leader or target at runtime so other agents can react. Unlike Unity tags there is no limit on their count, and they are queryable spatially.High performance and optimizable: No heavy APIs, low memory use, optional Burst for influence maps. You choose how often each part runs per agent, from target finding to score calculation, so you can tune it for your game. Want far away agents to think less often and change targets less often? Change two values on the Brain component based on their distance to the player.Full source code: We don't know all of your project's needs, so the asset ships with all source code, including the editor code and its UI Toolkit based UI.Easily extensible: No deep abstraction layers. Storage is plain ScriptableObjects, so everything can be read and modified by your own editor scripts outside the system. Classes are small, documented, and designed so a class's internals can change without breaking the code that uses it.Extensive documentation: A full manual which opens with the combat quickstart, genre cookbook pages, and complete API documentation.Support: Fast email and Discord support, and we take feature requests seriously. We also do custom development for your projects and can help you ease into utility AI.Parts of this package (demo content, documentation, and portions of the code) were developed with AI coding assistance. All of it is designed, reviewed, and tested by the developer, and the core framework is hand-written.




