Beyond the helpdesk - AI human collaboration frontiers ​
I recently read this article in Enterprise Times about how human/ai collaboration is key. I've been thinking about this topic for a while, and it's really cool to see how enterprises are actively working on this, to make the user experience and productivity amazing, and AI can do a lot to improve that.
🎤 World tour 💃 ​
The article does focus on the Salesforce World Tour which was recently in London, and I didn't know that it was a rock-concert-like actual world tour. But it definitely is, they're even coming to Australia in a couple of weeks 🤘
Salesforce's impressive statistic is Agentforce-resolving 84% of customer support queries with minimal human intervention-is a testament to the strides made in AI-driven automation. Only 2% required human intervention. Honestly these numbners are really impressive. But it represents just the initial phase of AI's potential in the enterprise.
The transition to AI-first operations is not just a trend (lol, genius). According to the Digital Enterprise 2025 report by Nasscom and Avasant, 27% of companies have AI in production or at scale, with another 31% at the proof-of-concept stage. Furthermore, 74% of enterprises expect AI spending to increase in 2025, underscoring the growing emphasis on AI integration in enterprises.
The thing that I think is missing from these kinds of presentations/reports (sorry enterprise, I promise I don't hate you), is what does that actually look like tomorrow?
Is it really just answering emails and customer support requests? I don't think anyone thinks that, but what does the future of human and AI collaboration look and feel like moving forward?
This is where the startup space can provide a bit of a window into that future.
🚀 The future ​
I think a good company to look at is Relevance AI in this space.
Relevance lets companies create a whole "AI workforce" (literally), each agent can spawn a series of sub-agents (in their new product Workforce). You can construct a whole network of agents to perform a tasks.
The reason you do this, is, just like humans, agents are much more effective when you can absract larger tasks into smaller, manageble pieces, and have a dedicated agent to just do that thing.
Most of the applications I've seen so far are in sales research, answering emails/queries. But it's actually configurable to the extent that you can imaging any application involving AI to be done autonomously. Including any external tools callable via MCP, by agents inside Relevance.
🤸 The human part ​
But there thing that's missing in that, is where is the human interaction? What does Human/AI interaction look like?
Right now, it's in interfaces like Cursor - you have a chat window, you type and it does stuff.
Companies like Gumnut (disclosure: my company, but you are on my blog...) are interesting in that - we have a real-time collaboration platforms where AI can type alongside you as a co-pilot. Think Google Docs, but the colleagues you're typing with are AI agents typing alongside you.
We'll see what Liveblocks) is cooking in this space too in the coming weeks.
There are lots of interface questions as to the best way to interact with AI agents as you type alongside them. Is it just another Cursor-like chatbot like TipTap have? Or something else?
We'll have a lot more to say on this soon 👀, but if you want to try it yourself (and you're in Sydney), come to our hackathon on July 2nd and see what you can come up with!