Reimagining IT with ChatGPT

Video authored by the channel shown, not by IANIA; all rights belong to the creator. Transcript and summary generated automatically; the original-language content is preserved.

Original language: English

Original video: Reimagining IT with ChatGPT · Channel: OpenAI · 03:06 · Watch on YouTube

Topics: AI agents, Task and workflow automation, Software engineering with AI

Summary

Kunal Malik describes OpenAI’s approach to reimagining IT by applying AI across individuals, teams, and the company, rather than automating isolated steps. He says ChatGPT has increased individual throughput by 5 to 10 times and helped make more employees builders of solutions. OpenAI’s IT agent provides immediate guidance, resolves some requests, or escalates them to human staff with context, while helping the help desk focus on complex and security-sensitive work. The team also uses AI to assess software needs and share reusable skills, with the goal of accelerating how companies realize value from AI.

Key points

AI should transform end-to-end processes, not just individual parts of them.
OpenAI considers AI’s impact at three levels: individuals, teams, and the company.
The speaker reports a 5 to 10 times increase in individual throughput using ChatGPT.
Employees closest to a problem can use AI to build effective solutions.
OpenAI’s IT agent offers immediate feedback, resolves some tasks, or escalates them to a human queue with context.
Automation lets human help desk teams focus on critical, security-sensitive, or uniquely human tasks.
ChatGPT helped assess software, identify risks and governance needs, and inform decisions about what to keep, retain, or defer.
The team estimates that existing software could meet 70% of software needs without new purchases.
Shared skills help teams reuse each other’s work and stay aligned.
CIOs can experiment, test, and deploy AI foundations to reach outcomes faster.

Chapters

Start	Chapter	Summary
00:01	Reimagining IT with AI	Kunal Malik outlines how AI can amplify individuals, teams, and the company, and describes the shift toward employees building solutions.
01:04	An AI agent for IT support	OpenAI’s IT agent provides immediate guidance, resolves requests, or escalates them with context while freeing human staff to handle critical work.
02:06	Assessing software needs	ChatGPT helps evaluate software and its risks, with the team estimating that existing tools can meet 70% of software needs.
02:36	Sharing skills and accelerating AI adoption	The IT team shares reusable skills across teams, while the speaker encourages CIOs to experiment, test, and deploy AI to achieve outcomes faster.

Quotes

We’ve seen 5 to 10x increase in throughput of individuals. — 00:32

everybody is becoming a builder. — 00:32

you get immediate feedback around what you can do to unblock yourself — 01:04

70% of the software needs can be met with existing software — 02:06

That is collective acceleration. — 03:06

Concepts and entities

Kunal Malik, OpenAI, ChatGPT, IT agent, ticket dispenser skill, SLA

Full transcript (original language)

00:01 [music] >> Hi, I’m Kunal Malik and I lead the global [music] IT team at OpenAI. Adding AI to just specific parts of the process is not going to be enough. To really reimagine the process completely [music] is the paradigm in which we can transform. We look at it in three layers. How we amplify individuals by giving them AI tools so that they can increase their throughput. [music] How we amplify an entire team, that’s where you have to go to the end-to-end workflow, and how you amplify the entire

00:32 company. [music] And what we’ve seen using ChatGPT work is amazing. We’ve seen 5 to 10x increase in throughput of individuals. The biggest operating model shift is that everybody is becoming a builder. And what’s interesting is the folks that are closest to the problem actually know how to create the best solutions [music] with AI. >> Historically, in IT organizations, whenever you have a problem, you come to a physical help desk location or you file a ticket with IT [music] in order for that to be addressed. Here at OpenAI, we’ve deployed IT agent. Rather than coming and waiting hours or

01:04 [music] days for a human to see your task and help you with it, you get immediate feedback [music] around what you can do to unblock yourself, or it’s escalated to a human queue with context, or the IT [music] agent itself will actually resolve that task for you. It also enables our human help desk teams to be able to focus on those really critical tasks that have high security needs [music] or are things that humans are really just uniquely positioned to do. >> I’ve seen a lot of our support team actually go and build their own micro sites, build their own [music]

01:34 agents to be able to help them do their jobs more efficiently. >> big processes within IT and transform them and really build that credibility to [music] understand how value can be created with AI. >> Software used to be governed and assessed at a point of time, and that limits its [music] ability to truly understand software for a fast-growing company. There are hundreds, if not thousands, of software [music] in any given company. The tool was able to first group them into roughly 30 categories. [music] ChatGPT work with able to do is give me a initial assessment of what [music]

02:06 software should be potentially keep and grow, and then looking at the risks and governance around them. It actually give me the first set of advice. A third you need to review, a third you need to retain, another third we need to defer. [music] With new process, what we think 70% of the software needs can be met with existing software >> [music] >> without buying new software. So, it’s very powerful analysis that traditionally I think I would take a couple [music] months to get to. >> Teams that are learning from use cases

02:36 can then take those learnings, be the starting point for other teams, making those teams further ahead. >> We created [music] the ticket dispenser skill that allows you to just dispense five tickets um based off of the highest priority or aging ticket based off of SLA breach. This is where we store all of our skills that [music] the IT support team has created. It’s a good way of sharing skills so that everyone stays in sync with the whole team and everyone’s using a fresh up-to-date

03:06 skill. >> CIOs at this point [music] have to make these big bets on how they want to create the right foundation for the company to unlock the value from AI. Having the ability to experiment, [music] to test, to deploy, that allows them to achieve the outcomes much quicker. That is collective acceleration. >> [music] >> We got to get some new ones in. >> Yeah, it’s nice.