ChatGPT integration released, looking for ideas!

ChatGPT is a revolutionary technology that could drastically improve productivity. MSPbots released the ChatGPT integration on 03/06. We are looking for new ideas and alpha testing. Please follow this screen to link your OpenAI API,
OpenAI Integration Setup

and provide your ideas about how to use OpenAI API by replying to this post.

1 )Another suggestion – re-write the ticket subject if under x characters or matches X
2) Assign company if catch-all if company data is in the body
3) Change the priority of the ticket based on the context of the ticket body |

  1. Check the sentiment of the notes and comments that is added by the client, write it to a dataset, then use the dataset to create a dashboard and bot to alert management
  2. Check the reply from the client, if they give the instruction to close the ticket, then auto closes the ticket.
  3. Give input windows in the Browser to compose a time entry or reply, polish the English to make sure there is no grammar or spelling error, and format the time log based on the company standard.
  4. Detect the type and subtype progressive, based on the original description and follow-up reply and time entries, and alert the ticket owner or dispatch to approval for the changes
  5. Write the ticket summary into the solution field when the ticket is ready to close.
1 Like

A heads up…“(2) Assign company if catch-all if company data is in the body” may give some security concerns if not thought out or used properly. This would potentially allow someone to easily impersonate a user at a company. Think about a password reset using a gmail account with the company’s signature.

When we have a legit catch-all from a customer we will verify the user though a second form of verification - such as a phone call to their main company number.

@Justin_Bennett , good point. These are very preliminary ideas, and this kind of feedback is very good.

We can create an MSPbots dataset instead of updating the PSA directly, then create an interface to ask for human intervention, with a simple “yes, please change”, or “no, don’t change” interface, it will not be fully automated, but still save a lot of time since AI is giving the suggestions already. Assume most of them are right, also, the negative feedback could be feedback to AI to reinforce learning for AI.

Offer potential solutions based on ticket content, however, this would greatly depend on the content -as usual garbage in… garbage out scenario.

Another thought since some ‘techs’ are lazy and don’t put anything of substance in notes - analyze the ticket content quality and request and update if the content is 'sub-par or of no use in resolving the issue again.

  • Ticketing processing
    ** Categorizing and labeling, the type, subtype, item, budget, priority, assignee etc.
  • Client reply sentiment, intention analysis, and automated actions like escalating, replying, or closing the tickets
  • Using AI to analyze phone calls and attaching them as time entries, greatly improving quality of service
    ** Using the GPT to answer the phone calls to answer basic questions
    ** Time entries review with wording, tonality, and compliance
  • Content generation
    ** RMM script generation
    ** Developing an incident response plan
    ** Writing job descriptions for playbooks for each position
    ** Marketing copywriting
  • AI-generated dashboards and bots
    ** Help data analysts and automation engineers speed up the BI and RPA development
  • Training AI with your own data so an AI knows your MSP more than anyone else
    ** Like ticket data, internal knowledge base, employee handbook etc.

From one of the clients:

Wants to know if there’s any integration between the AI tool and the next ticket feature. e.g. AI being able to look at previous tickets in autotask and ingest all of that information like the usual tickets being worked on by different techs and from there, leveraging all the data that it’s consumed to make more intelligent decisions moving forward about which tickets should be assigned to whom. 2nd, with all the information gathered by the AI, can the AI look at the content of the ticket and analyze to generate a response acknowledging receipt of the ticket and add standard suggestions their issues

Analytics - Is there a way to tie data together and provide intelligent recommendations based on those data

Is there progress on getting access to the Manage “discussion” thread to be able to do this?