AI Software Integration

[WelcomeSAC]

Hi everyone! Firstly, please forgive any inaccuracies in this, I am no expert!

I am currently looking into AI software and how we can integrate this into our systems to improve and/or automate many of our system processes.

I would be interested to hear what other people have found to be useful, or equally, not useful? Specifically, I would be looking at integrating to Dynamics 365 CRM to automate workflows which are currently being manually completed.

I have also been having a look at ChatGPT 4.0 and have so far been impressed. I am looking into the possibility of building my own GPT, has anyone done this so far and found it helpful? What are some limitations you have experienced with this?

Any and all feedback welcome, thank you!

 

[Ozzy]

I am currently looking into AI software and how we can integrate this into our systems to improve and/or automate many of our system processes.

My first comment would be to disregard the phrase 'AI' and focus on the workflow aspect of what you are trying to achieve. Document it out in flow charts and decision trees, because that isn't really AI but is just that - a workflow. Then once you have documented the workflow and all the decisions that could surface through that workflow, and the outcomes, you will then be better placed to build the systems around that.
We use Jira internally within my business for workflows. No AI involved and simply decision trees, which edge cases along the way where something may fall out the workflow a human picks it up.

I am looking into the possibility of building my own GPT, has anyone done this so far and found it helpful?

I have, I've built three and they are great. Solely for producing editorial with the correct tone of voice, I'm a big fan but the output still needs a human to look over it. You can tell how busy I am at any time in the editorial I publish on here, if it is less 'AI' robotic I've had some time on my hands to fine tune it and add some personalisation, but if it is so obviously AI then I've been a bit busy at that time .

Building a custom GPT is good though as you can pre-program it with your brand guidelines, tone of voice and things it needs to consider with each output saving you a lot of time in prompt building.

 

[rsshep]

Ozzy hit the nail in the head when he said that before you start mixing AI into your workflow, you really need to understand what you're working with. The key is making sure AI adds value - making things quicker, simpler, or cutting down on the boring stuff. Does it pay off in the long run?

AI worked for a client of mine. They're in recruitment, super competitive. Every day, their inbox is swamped with emails about job opening, the same ones that land in every other agency's inbox. The race is on to match those jobs with the right people. This process was a mess - important job details hidden in messy unstructured emails, a nightmare to search though, and just way too time-consuming.

What we did was, setup AI to dig through each email as it arrived, pull out the important bits, and organise everything in a way that's easy to search for through a web portal. Plus, we added a feature that lets their recruiters get alerts when an email matches what they're looking for. This way, they can jump on opportunities faster.

It's been a massive win for them. They were really struggling with the email chaos before, and nothing they tried was working for them.

 

[Vorcy1974]

Hi everyone! Firstly, please forgive any inaccuracies in this, I am no expert!

I am currently looking into AI software and how we can integrate this into our systems to improve and/or automate many of our system processes.

I would be interested to hear what other people have found to be useful, or equally, not useful? Specifically, I would be looking at integrating to Dynamics 365 CRM to automate workflows which are currently being manually completed.

I have also been having a look at ChatGPT 4.0 and have so far been impressed. I am looking into the possibility of building my own GPT, has anyone done this so far and found it helpful? What are some limitations you have experienced with this?



Any and all feedback welcome, thank you!

Hi,

Sounds like you’re on a really common path right now — a lot of teams are trying to plug AI into Dynamics 365 to reduce manual work.

From what I’ve seen, the most practical use cases aren’t full “AI automation of everything”, but small parts of the workflow, like:

sorting or tagging incoming leads
summarising emails or customer history
helping draft replies or next steps
pulling structured info from messy text

About building your own GPT — most people don’t actually build a “new model”. They usually just combine ChatGPT with their own data and system integrations. That tends to work much better in real business setups.

The main limitation is that AI is not great at long, fully automated processes where everything has to be 100% consistent and predictable. It works better as a helper inside existing workflows rather than replacing them completely.

In general, I’d start small with one or two CRM processes first, get that stable, and then expand from there.

 

[Data Swami]

As Ozzy has said disregard AI for now.

Why are you looking at Dynamics 365 or is this your CRM at the moment? What specifically are you looking to automate?

There are ways to automate in 365 with Power Automate etc but I will say that its not exactly straight forward for non techies and can take alot of time without the right expertise.

Ive built a multitude of different AI solutions and automated solutions there is space for both to work together but it has to actually make sense for it to work like that. For instance a Copilot agent I produced for a client where it was able to draft a full medical device survey within 3 minutes rather than the hours it previously took to get it to that stage. It made sense there as the AI was there to grab lots of information and then a separate automation to render the survey into a word doc with all the right formatting.

Give us a shout what you are currently spending the most amount of time on as that will help to pick out what you can look to automate first

 

[Keynote Speech]

My first comment would be to disregard the phrase 'AI' and focus on the workflow aspect of what you are trying to achieve.

I think that's a good point about starting with the workflow rather than the technology. It's easy to get excited about AI and automation, but if the underlying process is messy, all you're really doing is automating the mess.

A simple example might be customer enquiries. Instead of someone manually sorting emails, AI could categorise them, identify the common questions and route them to the right person, while the actual decisions and customer interactions still stay with a human.

 

[Paul Kelly ICHYB]

AI could categorise them, identify the common questions and route them to the right person

Do you need AI for that? Standard automation could probably do it, depending on the system you use - even email systems like gmail can do this.

 

[Keynote Speech]

Do you need AI for that? Standard automation could probably do it, depending on the system you use - even email systems like gmail can do this.

Absolutely right! I suppose the point where AI becomes more useful is when you start dealing with less structured inputs or decisions. For example, understanding the intent behind an email, categorising unusual requests or drafting a response rather than simply routing it. But AI isn't necessary for all businesses or workflows.