Loomy Tutorial: How to Use It - Multi-Agent Collaboration for Desktop Automation
From "one person doing the work" to "a team providing services," pets working for others automatically earn points, and the organizational structure of desktop intelligent agents is evolving.
From "one person doing the work" to "a team providing services," pet working earns points.automaticRecover health, desktopintelligentbodyOrganizational forms are evolving
Most users are usingAI AgentOften, one problem remains unresolved: complex tasks frequently require the collaboration of multiple professional skills, one...AIWhat if I can't figure it out?
This Loomy update provides the answer:"Partner groups" and "pet jobs" Two major features were launched simultaneously.
After using it for a few days, my impressions are:I'm finally starting to seriously consider the questions of "how to get things done with multiple people" and "how to keep users wanting to use it."。
This article starts from a truepracticalFrom a user's perspective, let's discuss how these two features are, what their drawbacks are, and my thoughts on the desktop.intelligentbodySome new observations on the direction.
I. Teammates: LetAIWork in teams, cooperate like a real team.
1.1 Let me first explain what this is.
SimpleIn other words, a partner is someone with a specific expertise.AIAssistants and collaborators are groups of people with expertise who work together to complete complex tasks.
For example, these are the pre-selected partner groups by the official team:
|
Team
|
Membership
|
What is it used for?
|
|
Content Growth Team
|
Six people including content strategists, research editors, and lead content writers.
|
Content creation across the entire chain, from topic selection to distribution.
|
|
Product Launch Team
|
User researchers, product managers, interaction designers, etc.
|
The entire process from user research to product launch
|
|
Business Solutions Team
|
Industry researchers, solutions architects, financial forecasting consultants, etc.
|
Industry solutions + financial forecasting
|
|
Data Management and Analysis Team
|
Data Analyst, Financial Analyst, Data Visualizer
|
Turn operational data into business recommendations
|
It sounds impressive, but I actually tested it in a scenario.
My testI've recently been considering the direction of a personal project, and I'd like to do something...SimpleLet's try out the product solution. I called on the "product release team" and gave them a very vague requirement: "I want to create a small tool to help programmers record technical notes. Help me think of how to do it."
- User ResearcherFirst, a "target user profile" (tech blog authors, frequent GitHub users, etc.) was generated.
- Product ManagerBased on the profile, a PRD draft was produced, including core functional modules.
- Interaction DesignerA preliminary page structure suggestion was provided.
- Market CommunicationI've added a "Cold Start Promotion Strategy" document.
The whole process took about three to five minutes, and the quality of the output wasn't exactly "amazing," but...The structure and logic are complete.Indeed, it transformed "a vague idea" into "a preliminary, discussable framework for a solution."
My judgmentIf you're starting a project or writing a proposal from scratch, a team of partners can help you.Build a framework, create an outline, and fill in any gaps in your knowledge.The efficiency improvement is real. But if you are already an expert in the field, its output may only be a "passing grade" and you will need to spend more time refining it.
1.2 From a technical point of view, this thing is quite interesting.
While there are still some rough edges in the user experience, from a technical architecture perspective, the team has several noteworthy designs:
First, role division relies on the System PromptconstraintEach partner is pre-defined with specific role commands and skill limits, essentially in...AgentThe framework layer implements role isolation. This differs from "using a single, generic model for everything" in that...Each partner's output style and thought framework are fixed and will not deviate from the expected path.。
Second, task routing is a critical engineering point.You give the team a task, and it needs...automaticDisassemble and distribute to the corresponding partners. Behind this is a...Task planning and schedulingThe logic of which partner should work first and which partner should wait until others have finished before starting needs to be arranged.
Third, moreintelligentbodyCollaborative product packaging. manyAgentCollaboration in academia andopen sourceThe community has discussed this extensively, but Loomy has made it a reality.Product features that work right out of the boxFor developers focusing on this area, the team-building group is a good example to observe. It allows us to see "the ideal..."Agent"Collaboration" and "productization"AgentWhat trade-offs are involved in "collaboration"?
1.3 and SingleAgentWhat are the differences between the two models?
In short:oneAgentYou ask a question and give an answer; the team is you who give a goal, help you organize an "online meeting" and deliver the meeting results.
II. Pets Working: An Underrated "Humanistic" Design
If team-up is an "upgrade in ability," then I think working with pets is an "upgrade in experience."
2.1 How does it work? You don't have to do anything; it earns money on its own.
The logic of pet working is specialSimple:Turn on the pet feature and use Loomy normally. Your desktop pet will occasionally pop up and say, "I've earned XX points for you," and then the points will be credited to your account.。
You don't need to do anything extra during the entire process:
- No need to sign in, no need to do tasks, no need to worry.
- It will "sneak in and do the work" while you're at work.
- They'll come and report back to you once they've earned points.
- You can check every single income transaction on the points details page.
I used Loomy for three days, handling some tasks normally each day, and accumulated about three or four "passive" points, the amount of which was not large.Every time my pet jumps out and says, "Master, I've earned you points!", I do get a "Wow, that's great!" feeling.。
2.2 Why I think this design is quite clever
Previously, Loomy's points system required users to actively spend their points by performing complex tasks or making calls.AIImage recognition and other tasks will deduct points. Under this model, users will have a sense of…Using points is like spending money.The psychological burden of "".
The roles of pets working have been reversed.During normal use, your points will...automaticincreaseThe significance of this shift lies in:
- Lowering the psychological thresholdUsers no longer feel that "using Loomy = burning points," but instead think, "Since my pet is helping me earn points while I'm using it, it's okay to use it more."
- passive motivationIn psychology,Variable ratio reinforcementThe uncertainty of when a reward will be received is more effective at maintaining behavioral frequency than a fixed reward. A pet unexpectedly popping up to give points creates more anticipation than simply giving a fixed number of points for daily check-ins.
- Emotional valueThe narrative of "Your pet is working for you" is much more heartwarming than "The system rewards you with XX points."
2.3 But to be honest, I also have some complaints.
The current problem with pet working is that...I can't see what my pet is doing.It says it went to work, but I don't know where it went, what it did, or how it earned points. If a visual interface like a "Pet Work Log" could be added in the future (e.g., "I went to the database today to help organize files and earned 10 points"), the experience would be much more complete.
Furthermore, the efficiency of earning points through pet work currently seems to be linked to usage frequency, but the specific rules are unclear. As a tech user who prefers "predictable feedback," I hope to see a more transparent system.SimpleEven a simple explanation of the rules, such as "calculated based on usage time and task complexity," would be acceptable.
Okay, now that I've said that, let me share my thoughts.Not enoughLocation:
- The partner's professional depth is limitedI tested several partners with strong industry attributes (such as the supervision and inspection partner). They performed well for shallow needs ("Help me break down the key points of this project"), but for deeper needs ("Help me analyze the potential risks of this project and give suggestions on how to avoid them"), the output granularity was still relatively coarse.
- The value and consumption rate of points earned through pet work need to be balanced.Currently, the rate at which points are earned and spent seems disproportionate—complex tasks consume points quickly, but pet work accumulates points relatively slowly. If this "spending more than earning" situation persists, the incentive effect of gamification will be diminished.
- The preset templates for team-building are sufficient, but could developers define their own team-building options?Currently, the official website provides four preset teams, but theoretically, the greatest potential of this feature is—Users form their own groupsFor example, I might want to assemble an "independent developer team" (product + design + development + operations), filling each role with my own defined skills. If SkillHub opens up the ability to customize team compositions in the future, the value of this feature will increase several times over.
IV. Conclusion: From "Tool" to "Organization," Loomy is forging a new path.
My overall assessment of this Loomy update is as follows:It's evolving from "a tool to help you get things done" into "a virtual team that you can call upon."。
Teamwork addresses the issue of "breadth of skills," as complex tasks cannot be handled by one person alone and require the cooperation of multiple roles.
Pet working addresses the issue of "willingness to use," allowing users to continue using the service without psychological burden, and even forming a habit through "incentives."
If you also have Loomy installed, I suggest you try the "Product Release Team" or "Data Management and Analysis Team".
Give you a real task that's giving you a headache, let it help you build a framework, and then see what works and what needs to be changed.。
The correct way to use a team is not to "let it do the work for you," but rather to "let it help you."fastFrom 0 to 60 points, you then refine it from 60 points to 100 points.