Jean Laval Chue HimDirector| Stella Aurorae Accountants Pty LtdSydney, Nsw, Australia
Hi all, I have been using AI co-pilots heavily lately but I find that for critical systems we cannot rely on vibe coding or AI Copilots. They do not understand the context or what I am trying to achieve. Software processes, like analysis, design, software development contextual application project management and functional and procedural programming are very hard for AI co-pilots and AI in general. However AI I find can be helpful for help on syntax. The result of using AI-Copilots is bits and pieces of code that are extremely difficult to maintain. And I find that AI makes my creativity and problem-solving skills decrease. My mind becomes lazy with AI co-pilots. But with Google search used together with AI, I can explore more quickly questions that I have and still make my own solutions with my creativity and problem-solving skills. However with AI I cannot reflect my originality and creativity. Although I must say AI has access to a vast knowledge base from which it can retrieve most probable answers to all previous questions asked before.
I would be curious to hear from Software Project managers and Research and Development Project managers who have used AI intensively and what are the results.
AI Product Manager (B2B) for PMI| Project Management Institute (PMI)Middletown, DE, United States
Hi Jean Laval Chue Him - what a great question. I'm not a software PROJECT manager, but rather a PRODUCT manager. I find myself using multiple AI platforms and tools to assist with the day-to-day tasks and find it a challenge sometimes when you cannot put all the context in (either because your organization hasn't approved the tool/platform or perhaps you have to take a chunk of time to sanitize the info which means diluting the value you get out of it).
I appreciate your take on using Google search with AI to explore your questions more quickly. I find myself using tools that are specific to the task (for instance ChatPRD for a product requirements document, Perplexity for research, and PMI Infinity for organizing, creating my schedules, and identifying risks). So I'm definitely stringing functions and outputs together, but it's accelerating my work and gives me time to focus more on the strategic piece of my role and working with customers. I think you're going to see more specialized agents (or copilots) and platforms that are niche and perhaps specific to the industry like PMI Infinity is. More platforms are offering secure and private options (building for the B2B/enterprise market) so that you can use your organizational documentation and artifacts, providing it with more specific context to your projects or programs. That's sometimes the drawback of platform that hasn't been approved by an organization (per our thought leadership research, organizations are a bit lagging behind in adoption because of data, security, and privacy risks). We're building out that data, security and privacy piece for Infinity right now for this very reason.
Are you using organizationally approved platforms or do you use your own tools?
...
1 reply by Jean Laval Chue Him
May 20, 2025 5:03 PM
Jean Laval Chue Him
...
Hi Kerry, I own my own business so I choose my own tools. For the time being I use Github copiolt, Perplexity and Google search engine, particularly LLMs are text based so I need Google search to get images.
I use AI to generate ideas but I still do use many manual project management processes checked and compared with AI suggestions.Because I am mostly in Finance and Software development, I am accountable for the work I produce and held responsible as I am a Certified Professional of the Australian Computer Society and a Certified member of the Institute of Public Accountants Australia I am bound by duty of care.
I am hoping to use more proven AI tools with time.
Cheers!
Jean Laval
Copilot isn't quite there for us. One of my devs tried to use it, last night, to help with a bug fix during a deployment. It didn't work and wasn't even close.
There are other cases where it has helped, and it's not the only AI tool we're trying out. We were going to check out GitHub Copilot, but only one of our devs is using tools that can leverage everything it offers, so we're trying CodeAnt, instead. It does well with small and medium size releases, but hasn't done well with large releases like our website software engine update. Not a great time for hallucinations.
For the AI,that is, not me.
...
1 reply by Jean Laval Chue Him
May 20, 2025 5:04 PM
Jean Laval Chue Him
...
Hi Aaron, yes your experience with AI is similar to mine.
Thanks for sharing.
Jean Laval
Saving Changes...
Jean Laval Chue HimDirector| Stella Aurorae Accountants Pty LtdSydney, Nsw, Australia
May 20, 2025 9:39 AM
Replying to Kerry Brooks
...
Hi Jean Laval Chue Him - what a great question. I'm not a software PROJECT manager, but rather a PRODUCT manager. I find myself using multiple AI platforms and tools to assist with the day-to-day tasks and find it a challenge sometimes when you cannot put all the context in (either because your organization hasn't approved the tool/platform or perhaps you have to take a chunk of time to sanitize the info which means diluting the value you get out of it).
I appreciate your take on using Google search with AI to explore your questions more quickly. I find myself using tools that are specific to the task (for instance ChatPRD for a product requirements document, Perplexity for research, and PMI Infinity for organizing, creating my schedules, and identifying risks). So I'm definitely stringing functions and outputs together, but it's accelerating my work and gives me time to focus more on the strategic piece of my role and working with customers. I think you're going to see more specialized agents (or copilots) and platforms that are niche and perhaps specific to the industry like PMI Infinity is. More platforms are offering secure and private options (building for the B2B/enterprise market) so that you can use your organizational documentation and artifacts, providing it with more specific context to your projects or programs. That's sometimes the drawback of platform that hasn't been approved by an organization (per our thought leadership research, organizations are a bit lagging behind in adoption because of data, security, and privacy risks). We're building out that data, security and privacy piece for Infinity right now for this very reason.
Are you using organizationally approved platforms or do you use your own tools?
Hi Kerry, I own my own business so I choose my own tools. For the time being I use Github copiolt, Perplexity and Google search engine, particularly LLMs are text based so I need Google search to get images.
I use AI to generate ideas but I still do use many manual project management processes checked and compared with AI suggestions.Because I am mostly in Finance and Software development, I am accountable for the work I produce and held responsible as I am a Certified Professional of the Australian Computer Society and a Certified member of the Institute of Public Accountants Australia I am bound by duty of care.
I am hoping to use more proven AI tools with time.
Cheers!
Jean Laval Saving Changes...
Jean Laval Chue HimDirector| Stella Aurorae Accountants Pty LtdSydney, Nsw, Australia
May 20, 2025 10:51 AM
Replying to Aaron Porter
...
Copilot isn't quite there for us. One of my devs tried to use it, last night, to help with a bug fix during a deployment. It didn't work and wasn't even close.
There are other cases where it has helped, and it's not the only AI tool we're trying out. We were going to check out GitHub Copilot, but only one of our devs is using tools that can leverage everything it offers, so we're trying CodeAnt, instead. It does well with small and medium size releases, but hasn't done well with large releases like our website software engine update. Not a great time for hallucinations.
For the AI,that is, not me.
Hi Aaron, yes your experience with AI is similar to mine.
Consultant| Canarys Automation LtdBangalore, Karnataka, India
This is a thoughtful and timely question—thank you for sharing it. I’ve had the opportunity to work closely with both development teams and stakeholders in AI-integrated environments, and I resonate with many of the points you've raised.
From my experience, AI copilots are powerful enablers, but they’re not replacements—especially when it comes to critical systems where context, architecture understanding, and business logic are non-negotiable. I’ve seen AI add real value in areas like:
Syntax suggestions
Boilerplate code generation
Speeding up prototyping or documentation
But when it comes to designing maintainable, secure, and domain-aligned solutions, human expertise still plays the central role.
You brought up a critical point—the risk of dependency. If developers lean too heavily on AI suggestions without reflecting or refining, it can blunt their problem-solving instincts over time. I encourage teams I work with to treat AI like a junior assistant or brainstorming tool, not a senior architect. It’s most effective when it helps accelerate thinking—not replace it.
As project managers and technical leaders, I think our role is to set the guardrails—deciding where AI is allowed to help and where human oversight must prevail. We also need to encourage developers to review, reflect, and refactor, keeping their skills sharp while benefiting from AI’s strengths.
...
1 reply by Jean Laval Chue Him
May 22, 2025 6:02 PM
Jean Laval Chue Him
...
Hi Ashwin,
thanks for your insights and relating your experience.
I believe AI could be the next topic for PMI to explore to produce guidelines for project management using AI in specific fields.
Insightful points, Jean. AI copilots are great for syntax and speed, but critical systems need human judgment, domain context, and deep design thinking. They’re tools, not replacements. The key is balance—using AI to enhance, not replace, our creativity and expertise.
...
1 reply by Jean Laval Chue Him
May 22, 2025 6:02 PM
Jean Laval Chue Him
...
Great point Pavan
Saving Changes...
Jean Laval Chue HimDirector| Stella Aurorae Accountants Pty LtdSydney, Nsw, Australia
May 20, 2025 11:07 PM
Replying to Ashwin Kumar H M
...
This is a thoughtful and timely question—thank you for sharing it. I’ve had the opportunity to work closely with both development teams and stakeholders in AI-integrated environments, and I resonate with many of the points you've raised.
From my experience, AI copilots are powerful enablers, but they’re not replacements—especially when it comes to critical systems where context, architecture understanding, and business logic are non-negotiable. I’ve seen AI add real value in areas like:
Syntax suggestions
Boilerplate code generation
Speeding up prototyping or documentation
But when it comes to designing maintainable, secure, and domain-aligned solutions, human expertise still plays the central role.
You brought up a critical point—the risk of dependency. If developers lean too heavily on AI suggestions without reflecting or refining, it can blunt their problem-solving instincts over time. I encourage teams I work with to treat AI like a junior assistant or brainstorming tool, not a senior architect. It’s most effective when it helps accelerate thinking—not replace it.
As project managers and technical leaders, I think our role is to set the guardrails—deciding where AI is allowed to help and where human oversight must prevail. We also need to encourage developers to review, reflect, and refactor, keeping their skills sharp while benefiting from AI’s strengths.
Hi Ashwin,
thanks for your insights and relating your experience.
I believe AI could be the next topic for PMI to explore to produce guidelines for project management using AI in specific fields.
Cheers! Saving Changes...
Jean Laval Chue HimDirector| Stella Aurorae Accountants Pty LtdSydney, Nsw, Australia
May 21, 2025 10:46 AM
Replying to Pavan Maddi
...
Insightful points, Jean. AI copilots are great for syntax and speed, but critical systems need human judgment, domain context, and deep design thinking. They’re tools, not replacements. The key is balance—using AI to enhance, not replace, our creativity and expertise.
No, AI copilots can’t fully replace software developers for critical systems. They assist with coding, debugging, and documentation, but human expertise is essential for architecture, security, ethical decisions, and handling complex, unpredictable scenarios. AI enhances productivity, but accountability and critical judgment remain human responsibilities.
PMI Team Member
Community Champion
