AI Features (MCP server for integration with ChatGPT)
I find myself more and more having Eramba open in a browser tab and ChatGPT in another asking stuff about how we do things, getting inspiration. As I recently tested a few MCP servers I searched to see if there was already one available and found that roadmap post.
I want to add that I am very much looking forward to that as that would give AI some context for my questions and I wouldn’t spend so much time copy/pasting stuff around.
Btw. there are some solutions which can create a MCP server based on API specs, have you guys looked into that already? It might be easier than writing one from scratch?
Has anyone in the community explored any of these solutions by any chance?
I need to revisit my brewing on this, but things have been busy.
I created a n8n node based on the API swagger that lets me do things within n8n and eramba data (you can also do this with the straight up HTTP Request node per API call). It should be fairly easy from there by adding a MCP Server Trigger that connects to workflows that gets the information you need.
mcp’s work best with direct data access (db’s) in our experience, that is what we have in place already for almost a year. their usage in real life is still debatable in our opinion. the example below is asking chatgpt to make a chart with risks per business units.
we are exploring as well agents with user defined instruction sets for specific tasks such as compliance reviews, online assessments reviews, etc. we will see, i would not trust yet an llm with these tasks without supervision.
I just wanted to add that I am not looking to let AI lose on our GRC, but it would be helpful if I can ask AI: have a look at my risks and see if you spot any overlaps we can consolidate as an example.
It could simply query all the risks but their details without me having to copy and paste all the info I think it needs to give me a qualified answer.