Can someone explain the concept of multibot architecture, When and how to use it?
Doubts -
Dear @Vaishnav_Tej ,
Multibot architecture is a framework in which mutiple robots working on queue/transaction items at a same time which increases speed of transaction processing.
If you have 10000 items to process & each item taking 1 minutes, then in normal RPA process(single bot). It will take 1000 minutes to process all items (1 item/ 1robot/1min).
Now, if you wanted to reduce processing time you can deploy two robots. Each robot takes one item at a time & whole process complets in just 500 minutes.(2item/2 bots/ 1min)
Now coming to your question:
With one bot how many process can be run in same VM or multiple VMs and at same time or different timings?
Ans: Only one process if its unattended mode.
If 2 bots are working on same process, how the work load is divided (Backend logic on queues/jobs/schedules) ?
Ans:
a. if you are using queue, no need to worry. Each robot automatically picksup item from queue. You dont have to configure any additional settings.
b. If you are using SharedExcel/DataTable, then you have to implement mechanism to stop duplicate processing. After 1st robot pick up item, mark it as processed in excel column so when 2nd robot trying to read it will first check status, if its empty then only it will pick the item.
How do you decide no. of bots required for automating a process?
Ans: It totally based on volumn & how fast business wanted to process the items. If single bot taking 1 hours to process all items & business wanted to complete to complete it in half hour, then you have to use 2 robots, If business wanted to complete process in 20 minutes, then you have to use 3 robots.
Hope this solves your queries, if yes please mark as solution 
Br,
NikZ
Thank you for the clarity .
one more doubt, as you said multibot architecture is a framework , is it like Re Framework ?
What is the process to implement multibot architecture ?
Or is it just using more no. of robots based on business requirement?
What is Multi-Bot Architecture?
Multi-bot architecture means using multiple UiPath robots to work together, either in parallel or at different times, to handle large volumes or speed up processing. This is typically used when there’s a high volume of work, strict deadlines, or a need to scale the automation.
A single bot can run only one process at a time. If it’s installed on one VM, only one process will run at any given time on that bot. However, you can schedule different processes at different times on the same bot.
If you have multiple bots on different VMs, they can run different processes at the same time. Also, with high-density robots, you can run multiple bots on the same machine under different user sessions, but again – one process per bot at a time.
When two or more bots work on the same process, the ideal way to divide work is by using queues in Orchestrator. You push all input data (transactions) into a queue, and each bot picks one item at a time. UiPath handles this logic automatically – it locks the item once a bot picks it, so there’s no overlap.
So both bots run the same process but on different queue items. No need to manually divide the workload.
This depends on three main factors:
You can estimate the number of bots using this formula:
Number of bots = (Total Volume × Average Time per Item) ÷ Available Time Window
For example:
If there are 10,000 items to be processed daily, each takes 30 seconds, and you want it done in 4 hours (14,400 seconds):
Total time needed = 10,000 × 30 = 300,000 seconds
One bot can handle = 14,400 seconds
So bots needed = 300,000 ÷ 14,400 ≈ 21 bots
Always add a buffer to account for retries or unexpected delays.
If this fulfil you needful then please mark it as SOLUTON
Cheers
Dear @Vaishnav_Tej,
Sorry if Framework word confuses you.
It a implementation method, not a framework.
What is the process to implement multibot architecture ?
Ans: Just deploy your process twice if you want 2 bots, trice if you want 3 bots to process items.
Bonus tip:
If your process include login into application, check if at a time application allows multiple login or not. Otherwise automation fails.
If only single login allowed, then you have just create copy of your process & use the new user for 2nd process.
Br,
NikZ
UiPath offers the REFramework (Robotic Enterprise Framework) as a template for your processes, designed around the Dispatcher/Performer mindset.
Here the Dispatcher(s) (bots) will populate an Orchestrator Queue (typically) and Performer(s) (bots) can consume and process the items in the queue.
This makes it easier to scale as you can simply wind up more Performers as needed.
I dont believe there is any set formula for defining how many Performers you would need, and I think it depends on your use case and the importance of deadlines.
Regards
Soren
Let me put down in simple terms
One bot at one time can run only one process …this is applicable to any unattended licnese..but at different times multiple processes can be triggered which runs one after another
If two bots are already started then jobs and schedules dont come into picture as they are already started. Coming to queues..the concept of queue itself is like picking one item at a time based on first in first out or based on priority,deadline etc..so two processed will pick one item at once and process each untill all are completed
It totally depends on how much load you ahve and the time you have to complete it..if load is high and time is less we can use more bots so that all bots run together and complete more items at once
Cheers
Hi @Vaishnav_Tej
Pls check this might be helpful,
One bot can run only one process at a time on a VM; to run multiple processes simultaneously, you need multiple bots or multiple VMs.
When two bots work on the same process, the workload is divided using Orchestrator queues where each bot processes different queue items independently.
The number of bots required depends on the volume of work, concurrency needs, and SLA requirements for the process.
Multibot architecture allows multiple bots to work in parallel, managed via an orchestrator.
Processes per bot: One bot can run one process at a time; for parallel runs, you need multiple bots or VMs.
Workload division: The orchestrator uses queues to distribute tasks so each bot picks up the next available job without duplication.
Number of bots: Calculate based on workload volume, processing time per item, available bot hours, and include a buffer for peak loads.
It’s best used when you need faster processing, handle large volumes, or require redundancy.