New: local AI, live debugger, replay, and automatic recoveryExplore the intelligence suite
Local visual AI for RuneLite

Automation that
understands what it sees.

Describe your goal to the AI, review the editable tree it creates, and let Automata Studio recognize images, objects, and entities while coordinating decisions, loops, and parallel monitors.

Local or OpenAI drafts Local perception Safe dry runs
Automata Studio
Engine active
MAIN FLOWSmart combat
ENTITYAmmonite CrabName recognized · 98%
VERIFYAttack confirmedHover + revalidation
IF IT FAILSVisual fallbackTemplates or ONNX
Health42%
Eat food
Prayer27%
Drink potion
12:41:08

Entity detected · interaction confirmed

LIVE
AIDraft generated
ENTITYName recognized
3modes: image, object, and entity
20visual references per step
200steps in each automation
20saved editable automations
01 / RECOGNITION

Three ways to see.
One way to decide.

Automata Studio chooses the right strategy for what you need to observe: an appearance, an interactive element, or a character with an identity.

01
IMAGE

Compare what appears

Search one or more references by similarity. Ideal for messages, icons, buttons, and states that must appear or disappear.

Bank interface visible
02
OBJECT

Find what you can use

Recognize weapons, food, potions, and interface elements through references or ONNX, including items inside the inventory.

Weapon · inventory slot
Ammonite CrabName locatedHoverAttack confirmed
02 / APPLIED INTELLIGENCE

Describe it.
AI builds the draft.

Write what you want to automate and choose free local AI through Ollama or your own OpenAI API. The copilot creates an editable tree; the local engine observes, decides, and acts.

Choose local AI or OpenAI inside the application
01

Observe

Combine images, objects, named entities, bar sensors, inventory, and OCR.

Local multimodal perception
03

Act

Move, wait, click, type, and hold keys with parameters that remain under your control.

Coordinated actions
AI where it adds value.

The copilot creates the plan once and always leaves it editable. Local AI can work through Ollama without an API key. During execution, screenshots, perception, and decisions stay on your computer.

Read frequently asked questions
03 / CAPABILITIES

More than an autoclicker.
A system that recognizes context.

Every tool is designed to build automations that respond to what is actually happening on screen.

Semantic vision

Entities recognized by name

Locate NPCs and enemies by their visible name, calculate the click point, and confirm interactions such as Attack before acting.

Included in Automata Studio
Local perception

Images and objects

Add up to 20 references per step or use a local ONNX model to recognize the game interface, inventory, and objects.

Included in Automata Studio
Game state

Sensors and OCR

Read health, prayer, inventory, chat, and menus; save their values as variables and react when they change.

Included in Automata Studio
Multitasking

Parallel monitors

Keep combat running while other automations watch health, prayer, and events with independent priorities.

Included in Automata Studio
Visual logic

Branches, loops, and variables

Combine IF / ELSE, IF FAIL, WHILE / UNTIL, Break, Continue, calculations, cooldowns, and Go To in the same tree.

Included in Automata Studio
Reusable

Automations as functions

Call any saved automation with JSON arguments and mapped results, or start it in parallel. Later edits are reflected in every call.

Included in Automata Studio
Safe testing

Dry run and live debugger

Validate recognition and logic without clicking or typing. Inspect the active step, decisions, variables, and failures while it runs.

Included in Automata Studio
Repeatable diagnostics

Recording and visual replay

Record frames and execution events, then replay a session without RuneLite to reproduce behavior and refine the automation.

Included in Automata Studio
Resilient execution

Calibration and recovery

Scale regions for different resolutions, wait for RuneLite, detect stalled progress with a watchdog, and launch a global recovery routine.

Included in Automata Studio
04 / MULTITASKING

Three signals.
One coordinated response.

The main flow continues while independent monitors watch critical states. When a condition changes, each one runs its response without losing context.

Session / Smart combat
Active supervision
THREAD 01Combat
CONTINUOUS
Ammonite Crab
Attack confirmed
Status Running
MONITOR 02Health
CRITICAL
HP < 45%
Eat
Status Action triggered
MONITOR 03Prayer
WATCHING
Prayer < 30%
Drink potion
Status Ready
05 / THE DIFFERENCE

Macros repeat.
Automata Studio observes.

An autoclicker follows coordinates. Automata Studio waits for visual evidence, interprets the result, and changes its path when the state does not match.

  • Fully visual construction
  • Modular and reusable processes
  • Simultaneous responses to events
Capability AutomataMacro
Recognizes NPCs by visible name
Confirms interaction after hover
Reacts to several events at once
Conditional loops and calculations
Dry run, debugger, and replay
Watchdog and global recovery
No-code editor
06 / EARLY ACCESS

Build the automation.
Let it observe.

Automata Studio is entering its commercial stage. The license, launch price, and secure purchase option will be published here.

Windows application Semantic recognition Local AI option with no API key
LAUNCH EDITIONComing soon

Price and purchase link are being prepared.

Review before buying No charges or reservations are active yet.
07 / FREQUENTLY ASKED QUESTIONS

What matters,
without the fine print.

Clear answers about the visual engine, AI, and how it works.

Where does Automata Studio use AI?+

The optional copilot converts your prompt into an editable step tree. You can use free local AI through Ollama without a key, or connect your own OpenAI API. Perception, conditions, and execution then run locally.

What is the difference between image, object, and entity?+

Image compares visual references; object locates items such as weapons, food, or buttons in the game and interface; entity recognizes an NPC or enemy by its visible name and calculates a click below the label.

Do I need images to recognize an enemy?+

Not necessarily. With NPC Indicators enabled, Recognize entity can find the exact name without samples. Images and ONNX models remain available as visual fallback.

Do I need to know how to program?+

No. Build the flow by choosing actions, adding visual references, and connecting conditions from the visual editor.

Can it watch health and prayer while doing another task?+

Yes. Run monitors or automations in parallel to observe several states and trigger a different response when each condition is met.

Can I test an automation without clicking in the game?+

Yes. Dry run validates recognition, conditions, loops, and variables while suppressing mouse and keyboard input. The live debugger shows each decision and failure.

What happens if RuneLite is not ready or the automation gets stuck?+

Automata Studio can wait for the RuneLite window, detect a lack of progress with its watchdog, execute a global recovery routine, and retry the failed step.

Does recognition happen in the cloud?+

No. Perception and execution run locally. If you choose local AI, the prompt also stays on your computer; it is sent online only when you explicitly select OpenAI.

Does it guarantee that an account will not be sanctioned?+

No automation tool can guarantee that. Every user must understand the rules of the service where it is used and take responsibility for their decisions.