Three Key Themes: AI, AI, AI

...and what about workflows?

As we awake from the week and rub the sleep out of our eyes, I feel a bit like I just got back from the theme park. If I sit still for too long and close my eyes I might think I am still on the rollercoaster ride; two days live in London hosted by ServiceNow. First at CreatorCon (31st October) followed by the World Forum (1st November) 2023, these two days gave us some real gravitational pulls, lots of highs, a few lows, and quite a few twists and turns. I won't be too surprised if I am not the only one left feeling a bit giddy!

First, let me get back down to Earth

During CreatorCon I chose the Advanced Developer track and spent the day getting hands-on exposure as I learned more about GraphQL, Now Intelligence and UI Builder (workspaces).

Upstairs there was a UK Public Sector Forum which I understand may have been an inaugural event, well attended and very well received. I heard positive things!

After all the daytime activities were wrapped up, a soiree for drinks, and a little food flowed, the finale followed closely with the ServiceNow London Developer Meetup. This was hosted by Kopal & Russell, with special guests: Chuck & Pranav. This included a menti quiz, with 120 participants! Shout to Sam & Andy. The game of jeopardy was a lot of fun.

Day One: CreatorCon

The CreatorCon sessions I attended were spot on. I put this down to the quality of the content but also because of the way they were executed. Setup and instance activation steps were clear and everything was well organised. The resources included a lab guide to follow, and each host was extremely experienced while they worked through the material on screen at a steady pace. This consistency allowed us to deviate and play, while providing ample space for dialogue and feedback. This is one of the best ways for me to learn.

What did I build?

GraphQL really whet my appetite as I had tasted it before and now was ready for my main course. I got really excited with the working examples and opportunities to validate my understanding. I can definitely see how I will be looking to use it in my application designs in the future... *wipes the whiteboard*

Now Intelligence provides some of the building blocks for the future of AI, and this lab on Task Automation got us working through some of the areas where we can take full advantage of AI/ML today. We began training NLU models based on setting up intents, utterances and vocabulary patterns, as well as AI Search for knowledge and Q&A use cases. Virtual Agent conversations were able to filter conversations into the correct queue. Some of the jargon may be a challenge, but if you have ever looked at building apps for Alexa using SSML, it does make sense (to me). Getting over any initial cold water shock helps build confidence so we can continue to build more intelligence into our workflows. We definitely jumped in at the deep end and had a great splash about.

At last we reached the UI Builder Photobooth App Part 2 finale, which was a lot of fun and allowed us to gain understanding about how the photobooth app was constructed. The final product was available for everyone to play with outside, alongside the Lego etching CNC machines and the infamous Liquor Bot on the hack farm. You might have seen a post I made on here, where I was using my laptop to take selfies, with the app I built in the final session, adding to the fun!

So how does this all make work better?

I always love seeing how creative we can be with ServiceNow and allow that creativity to flow back into the workplace as we think outside of the box. This is one reason I enter Hackathons. CreatorCon is an amazing opportunity to learn, build and play surrounded by new and experienced professionals.

Every activity involved in CreatorCon can be tied back to something that has the opportunity to deliver value for customers as we build stronger, better, safer.

Day Two: World Forum

Generative AI has landed on the ServiceNow platform, taking the form of Now Assist. Drafting content, understanding the intentions of the user and summarising datasets - are the top 3 use cases that are possible today, in Vancouver release.

Use Case: Language Generation

I saw demos that covered Gen AI in the shape of text-to-code and incident closure notes. I was thinking... what about drafting policies, identifying new risks and starting to close gaps in the control framework? Or priming indicators based on control objectives. Sometimes we just need to get started ("don't get it right/write, get it written..."). We could also consider drafting stories and acceptance criteria, or consuming the catalog of Test Management to define Automated Testing opportunities. A lot to unpack here.

Use Case: Knowledge Synthesis

A real-life use-case I am facing right now that I think would be possible to solve within this next technological leap: the ability to ingest a SOC2 Report and use it to pre-populate a Vendor Risk Assessment based on the information provided in the report. ServiceNow's Document Intelligence is very good at understanding pro-forma data and structured content, but ingesting lengthy reports written by different auditors while they were armed with their poetic licence is a different beast entirely. I also think Sprint Planning is an area of interest here.

Use Case: Intent Understanding

What is it you really want to achieve? And how can we align our workflows more effectively, to answer the real question and get access to best information at the very start. We saw the potential to register and understand user sentiment, which could potentially trigger workflows. Also consider demand generation and the aforementioned story writing. And, did I mention Sprint Planning?

Abstract the Complexity

A big topic that came to life was the complexity of every organisation's technical landscape. Its no secret that ServiceNow has a distinct ability to integrate across multiple systems and this presents a unique opportunity for unified experiences.

When we want to connect multiple systems of records, we need to consider the payloads from confidentiality, availability and integrity perspective. Performance is key, and so when I think back to the GraphQL lab , I realise the need for us to plan ahead, thinking about total experience, what data payloads are going to be required and defining the schemas upfront will only add structural gains , which further supports and enables AI use cases and the preparation that underpins it. Start getting fit now. As the saying goes: train hard, fight easy.

Furthermore, the Reality of AI session talked about how DWP moved from a single supplier (EDS, where I cut my teeth in the IT industry) to a vast array of computational complexity. Making sense of that, managing the risk and security, and continuing to deliver growth and scale, is an enormous undertaking where ServiceNow is uniquely positioned as the platform of platforms.

Should we be scared of AI?

In the UK the first day of the AI Safety Summit was held on the same day as World Forum, with government and technology leaders from around the world. This resulted in the signing of The Bletchley Declaration , so I think we can agree this is a serious matter. With great power, comes... <?>

So lets consider some of the downsides to AI...

• Fatigue. They say the pace of change is always accelerating so it could be we are entering the "trough of disillusionment" earlier than we might expect. Some customers are looking for the actual wins today. "I'm sick of it!" Understandably it is difficult to invest in something that has yet to prove itself, and securing funding on a "leap of faith" may not always be possible, or indeed ideal. We can all start getting prepared right now, today, by doing something but it is difficult to move with confidence when none of us are sure we aren't running in the wrong direction. This is not the time to burn out.
• Regulation. Large Language Models (LLM) are being trained at scale - in domain-specific environments. This reduces the hallucinations that might exist in the public domain. "How will this work in a regulated way?" What are the rules as the paradigm shifts? Will we realise too late, and how will we reduce harm to individuals? Legislation is always trying to catch up, in most cases we can always be guided by a strong moral compass. Ethics aren't changing, only the context in which they are applied.
• Trust & Liability. Who is responsible for the truths and lies which AI returns? Like automated driving if someone gets hurt. And what if the machine is too accurate, and too honest with the wrong person? What kind of RBAC (Role Based Access Control) approach is needed or possible, when all the data is tokenised? "What about Privacy?" We saw a demo where HR articles were returned for policies on parental leave as an employee learned about their pregnancy. If we are judging a user's sentiment, is there a health & wellbeing concern? These use cases also give rise to automated profiling and secondary use/repurposing of data.
• Privacy By Design and by Default - this principle should be front and centre for all organisations as they adopt new technologies. We need to stop asking "What if?" and start asking "What when?"

AI is here, and ServiceNow are putting AI to work. What this really means, and where we take it - is up to all of us.

We can all be masters of our own destiny.

#PhilGoesDeep