Based on many peer discussions, seemed like GitHub Co-pilot was really the only true AI product teams were using internally. Any other tools people are finding value with on internal side?

4.4k viewscircle icon8 Comments
Sort by:
CTO in Services (non-Government)2 years ago

I had my tech department all trial Co-Pilot and ChatGPT and the conclusion we reached was they add very little value. My software engineering teams are skilled enough that it is faster for them to write the simple code themselves, and the questions they do need to ask of the LLM were nearly always too specific and advanced to receive a correct answer. The same was true of my ML and DS teams. The only team who benefited was my DevOps (who also do cloud infrastructure) - they found Co-Pilot/ChatGPT could efficiently produce AWS CDK scripts for them, but this is a specific case because those particular scripts tend to be formulaic and long. 

I did start a wide ranging trial of Google Duet, but paused it as the key features in the marketing that drew me in - namely the email auto summaries, and 'attend-for-me' in google meets - still aren't released.

Chief Technology Officer in Healthcare and Biotech2 years ago

Other companies are using their own AI tools internally.

For example, Google's GCP Vertex AI is being used by them to help make decisions around cloud architecture, to pinpoint faults, and converting complex legacy database code to open source platforms.

Another is Amazon which, of course, uses AWS AI tools to predict customer demand, and to promote products to people.

Director of Engineering in Services (non-Government)2 years ago

We are using ChatGPT internally for:
1. Creating summaries of longer documents
2. Creating marketing materials based on internal documents
3. Creating social media posts for various different channels (LinkedIn, Instagram, Facebook, Twitter)
4. Letter writing (yes that still happens)

Global Head of AI, Data & Analytics in Software2 years ago

Bard for research with citations
Chatgpt for basic q/a and text generation
Gpt4 with own RLHF for agents
Was midjourney for graphics, evaluating Dalle3 now
Synthesia for informative video

Senior Vice President, Engineering in Software2 years ago

Other than GitHub Copilot, we have seen developers using Tabnine and Codeium very often with a lot of success. If you are an IntelliJ fan, do check out their custom AI available in their tools like Rover and Fleet. 

ChatGPT has been a revolution and probably is as good as any other option available today. Feel free to checkout these tools though.

Lightbulb on1

Content you might like

IT is the decision maker 81-100% of the time19%

IT is the decision maker 61-80% of the time49%

IT is the decision maker 41-60% of the time22%

IT is the decision maker 21-40% of the time7%

IT is the decision maker 0-20% of the time1%

View Results

Strongly agree13%

Agree70%

Neither agree nor disagree12%

Disagree3%

Strongly disagree

View Results