Keith: Can you give me a specific example from IT support? Can you share an actual case where an ambient agent discovered something before the help desk did? Babin: Absolutely. Let's say there are 10 VPN disconnections in a region. Surrounding agents detect patterns and alert IT.
Difficult to program in advance due to different thresholds and triggers. Another example: Agents can direct employees to the appropriate DRI (directly responsible) to resolve issues based on past resolution patterns.
Alternatively, you can notify new managers about the security training they need to complete. Without this, you could be missing out. Keith: It's like a positive “nudge” that people respond to. It's more like a compliance reminder than an email with no response. Babin: That's right.
People respond better when reminders come from agents they already use. One customer increased compliance from 30% to 60% simply by switching from email reminders to ambient agents. Keith: Wow. That's a huge improvement. What about time-sensitive tasks like approvals? Babin: Great example.
One customer tracked software approvals. Before the agent was introduced, it took an average of 10 hours for a VP to approve. After introducing the environmental agent, that time was reduced to 9 minutes. It's not just about speed. It’s a change in behavior driven by context and helpful nudges. Keith: Yes.
In my experience, some of the slowest processes are approving things like vacation and expenses, which often get buried in the inbox. Bhavin: That's also a perfect use case. Agents track requests, notify managers, and ensure nothing is missed.
Keith: Are some industries more ready for ambient agents than others? Bhavin: It's less about the industry and more about the readiness of the company, the systems in place and the mindset of the leadership. We have been successful in the healthcare, defense, retail, pharmaceutical and technology sectors.
At the end of the day, it comes down to whether your company is moving forward with the cloud and has already invested in automation. Keith: Of course. But I can hear the IT guy asking, “What? “Is this just a tool that bolts onto an existing stack?” How easy is integration? Bhavin: Ambient agents are challenging the current model.
Today's copilots mostly power platforms like Salesforce and Microsoft. But ambient agents work across systems like Slack, Workday, Jira, and ServiceNow. It requires more extensive integration, but provides much more value. Fortunately, platforms are becoming more extensible, and that can help.
Keith: But it also raises privacy and security concerns. How do you manage trust and access? Bhavin: Privacy and autonomy are paramount. Agents must respect the same corporate rules already in place. “God Mode” access is prohibited. You cannot let the agent guess permissions.
Access should be deterministic based on roles, ACLs, and existing policies. Autonomy is acquired over time. We recommend starting with low-risk actions, such as resetting your password. Only later does the agent handle more sensitive tasks.
Keith: And once trust is built, users will be more willing to relinquish control. Users are more willing to relinquish control, especially if agents reassure them that access can be quickly restored. Babin: That's right. The key is to reduce friction.
People are more willing to let go of unused licenses if they know they can quickly get their tools back. It is important to change that way of thinking. Keith: Overall, ambient agents improve speed. And speed solves a lot of problems. Babin: Speed is the real game changer.
If you act quickly, mistakes will be corrected quickly. This agility makes your business more efficient and profitable, driving growth. This is why AI is driving a large portion of GDP growth. Keith: Do you think ambient agents will become standard within the next few years?
Bhavin: It's not standardized in the strict IT sense. However, communication protocols like MCP are becoming more common and can help LLMs access data. Some agent standards, such as ATOA, are still under development. The key is how easy it is to build useful and sophisticated agents. Competition is fierce there.
Keith: What about the day-to-day experience of workers? Will people be managing agents rather than doing their own work? Babin: That's possible. It's like conducting an orchestra of agents. But it requires maturation of systems and coordination. Not all roles change this way.
Many jobs involve complex decision-making, collaboration, and judgment. That said, over time, more and more “jobs within jobs” will be automated. Keith: And humans don't like to change overnight. So it gradually happens. Babin: Yes. It takes time and intentional development.
But we are at the beginning of a very exciting frontier. Keith: Babin, thank you for joining us today. This was an interesting conversation. BAVIN: Thanks for having me, Keith.
