Crewai crewai_orchestrator_workers_tutorial.ipynb

[animeshkr7]

So i was going through this file and it didn't work , though i am not using gpt , still have a high end llm but this workflow code does not work . can it actually even be possible to use specific agents and task according to my input when many agents and tasks are initialized in crew . dont provide me with the routing way.

[axiomofjoy]

Hey @animeshkr7, can you provide details on the expected behavior vs. what you are actually observing in the notebook?

[animeshkr7]

Thanks for replying , I got my issue solved it was just that I previously did some changes in crewai internal code myself and that affect the follow ... Reinstalling it solved the issue. But regarding the parallel processing code for crewai , I am still not able to understand it how come that sequential process act in parallel when we have ran it the same we do normally

…

On Tue, 27 May, 2025, 11:49 pm Xander Song, ***@***.***> wrote: Hey @animeshkr7 <https://github.com/animeshkr7>, can you provide details on the expected behavior vs. what you are actually observing in the notebook? — Reply to this email directly, view it on GitHub <#7720 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AGGUIJQYH43TFF2UJJEWIT33ASUCVAVCNFSM6AAAAAB565MLO6VHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTGMRYG4ZDAOA> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[axiomofjoy]

Hey @animeshkr7, glad you got it working. I'm not sure I follow the question, and afraid I am not an expert in CrewAI. It sounds like you were expecting the notebook to execute sequentially, but it is executing in parallel instead?

[SrilakshmiC]

Hi @animeshkr7, I'm glad you were able to resolve your issue!

About your question, when referring to the parallel processing code, are you referring to the parallelization notebook here? If that is what you are referring to, the Process.sequential setting in the CrewAI parallelization notebook might make it seem like the agents are running in parallel, but technically, they are not. So each Crew instance processes its assigned task one at a time in a sequential manner.

The illusion of parallelism comes more from how the notebook structures the workflow. It creates multiple independent Crew objects, each assigned to its own task (like in this case, different research topics). Although each crew runs sequentially within itself, nothing prevents us from executing these separate crews concurrently using standard Python tools like ThreadPoolExecutor or asyncio. Hence by doing so, we can achieve that true parallel execution, where multiple tasks are handled simultaneously across different agents.

TLDR: While CrewAI doesn't yet support parallel task execution within a single crew, you can still scale out concurrent processing across multiple crews — making it a practical and flexible approach for parallel workloads.

Hope this helps!

[animeshkr7]

That's where my question is . The code demonstrated only one crew running multiple different agents like reseacher 1 ,2 but all had to do same work with the difference being the topic , so how can we make them parallel , not the crew using python libraries but for that agent case .

…

On Fri, 30 May, 2025, 11:02 am SrilakshmiC, ***@***.***> wrote: Hi @animeshkr7 <https://github.com/animeshkr7>, I'm glad you were able to resolve your issue! About your question, when referring to the parallel processing code, are you referring to the parallelization notebook here <https://github.com/Arize-ai/phoenix/blob/main/tutorials/agents/crewai/crewai_%20parallelization_tutorial.ipynb>? If that is what you are referring to, the Process.sequential setting in the CrewAI parallelization notebook might make it seem like the agents are running in parallel, but technically, they are not. So each Crew instance processes its assigned task one at a time in a sequential manner. The illusion of parallelism comes more from how the notebook structures the workflow. It creates multiple independent Crew objects, each assigned to its own task (like in this case, different research topics). Although each crew runs sequentially within itself, nothing prevents us from executing these separate crews concurrently using standard Python tools like ThreadPoolExecutor or asyncio. Hence by doing so, we can achieve that true parallel execution, where multiple tasks are handled simultaneously across different agents. TLDR: While CrewAI doesn't yet support parallel task execution within a single crew, you can still scale out concurrent processing across multiple crews — making it a practical and flexible approach for parallel workloads. Hope this helps! — Reply to this email directly, view it on GitHub <#7720 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AGGUIJW2J7EZWJVC5JSMPQ33A7UNZAVCNFSM6AAAAAB565MLO6VHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTGMZRG43DQNI> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[animeshkr7]

Still as mentioned I can make those task as instead of being inside one single crew they can be with 3 different crew , but then again we that will make it manual for the summarizer agent as we have to mention to take all 3 summaries then process it On Fri, 30 May, 2025, 2:25 pm Animesh Kumar Singh, ***@***.***> wrote:

…

That's where my question is . The code demonstrated only one crew running multiple different agents like reseacher 1 ,2 but all had to do same work with the difference being the topic , so how can we make them parallel , not the crew using python libraries but for that agent case . On Fri, 30 May, 2025, 11:02 am SrilakshmiC, ***@***.***> wrote: > Hi @animeshkr7 <https://github.com/animeshkr7>, I'm glad you were able > to resolve your issue! > > About your question, when referring to the parallel processing code, are > you referring to the parallelization notebook here > <https://github.com/Arize-ai/phoenix/blob/main/tutorials/agents/crewai/crewai_%20parallelization_tutorial.ipynb>? > If that is what you are referring to, the Process.sequential setting in the > CrewAI parallelization notebook might make it seem like the agents are > running in parallel, but technically, they are not. So each Crew instance > processes its assigned task one at a time in a sequential manner. > > The illusion of parallelism comes more from how the notebook structures > the workflow. It creates multiple independent Crew objects, each assigned > to its own task (like in this case, different research topics). Although > each crew runs sequentially within itself, nothing prevents us from > executing these separate crews concurrently using standard Python tools > like ThreadPoolExecutor or asyncio. Hence by doing so, we can achieve that > true parallel execution, where multiple tasks are handled simultaneously > across different agents. > > TLDR: While CrewAI doesn't yet support parallel task execution within a > single crew, you can still scale out concurrent processing across multiple > crews — making it a practical and flexible approach for parallel workloads. > > Hope this helps! > > — > Reply to this email directly, view it on GitHub > <#7720 (reply in thread)>, > or unsubscribe > <https://github.com/notifications/unsubscribe-auth/AGGUIJW2J7EZWJVC5JSMPQ33A7UNZAVCNFSM6AAAAAB565MLO6VHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTGMZRG43DQNI> > . > You are receiving this because you were mentioned.Message ID: > ***@***.***> >