Join the event trusted by corporate leaders for almost two decades. VB Transforma connects people building AI Real Enterprise. Learn more
This Aimic AI age -moving age requires complete rethinking of how we build the software. The current API interfaces of the enterprise were built for operate by people; The API of the future will be multi-model, native interfaces.
“We need to build a type of API interfaces that will work well with agents, because agents are those who will now interact with the API, not people,” Merrin Kurien, Chief Engineer and architect of the AI platform in AI AT IntitutHe said during the breakfast of women in AI this year VB Transform.
Kurien conducted a lively discussion about the present and the future of AI agents with other AI Mai-Lan Tomsen Bukovec practices, leader of engineering and product in the field of storage and computing services in AWSand tiffany are, SVP product for the platform and enterprise at the address Atlassian.
“I would like to think in five years, agents will be the mainstream,” said Kurien. “Many challenges we face today will probably be overcome thanks to a better tool, if the last two and a half years are any clue. How will you be prepared? It depends on your investments.”
How an intutit receives invoices and AWS supports faster migration
The intutute uses agents and sees “amazing progress”, Kurien informed in the panel on the stage, which was moderated by Betsy Peretti, an innovation and design partner in Bain.
In particular, the platform company of financial technologies has registered automated generation of invoices and reminders to it Quickbooks An offer that is popular among petite and medium -sized companies (SMB).
“We saw the companies earn on average five days faster and exists by 10% more likely that the invoices receive full,” said Kurien.
AWS also recorded success AWS TransformTomsen Bukovec, agile infrastructure that migrates .NET, Mainframe and VMware loads.
Conventional migration scenario, as she described: the customer would go to the application owner and ask, for example, to the Windows application for an application based on Linux operating on AWS. “And guess what they would say?” Take the number. You are priority number 42. ”
But now enterprises can perform most of these migrations with the aid of AI. “Your generalistic teams are able to do much more work on their own and this reduces the request to a specialist,” said Tomsen Bukovec. “It changes migration as an industry.”
Ultimately, as AWS and others evolve, it will be closely related to how customers operate artificial intelligence, she said. She was surprised that the amazing progress in artificial intelligence “really makes us make a new look, a hot shot”, how to build applications.
“When we build agency infrastructure and include artificial intelligence into the mission of our companies, we not only accept technology and introduce it to work,” said Tomsen Bukovec. “We will actually change the nature of the workplace, workforce.”
She added: “We see it now. We see that this is happening at the speed of Warp.”
How Atlassian learns from internal experiments and with clients
Atlassian has a thoughtful approach to the inside of AI agents.
For example, the project management platform has launched an on -board agent to aid novel employees access all materials needed to start work. In the first month of the premiere, the agent issued 7,000 demands. Now it is an ordinary part of the implementation process, to say.
Meanwhile, the company market team has many interface points with clients, which can make it hard to collect all the necessary context. Atlassian built a client agent who connects all these data, and informed that this is one of his most popular agents, used by 80 teams in the whole company. “I use it a lot before I talk to customers,” she admitted.
In Atlassian, there is a mighty responsibility for “dog food” – using their own products and services – and ITeratively experiment to aid customers during evolution with AI. This work can then be translated into what Atlassian Ships on customers after removing from the box.
“Not only comes from engineering; he comes from the whole organization,” she said. “So what can you do programmatically to create the creativity of all interfunctional, combine ideas, designing work flows?”
The company recently introduced a “team work collection”, a selected selection of applications – Jira, Confluence and Loom – managed by “Rovo agents”. It is built into its platform and supports various aspects of the cooperation process. For example, before the meeting, the agent will be a “really nice summary” based on Confluence pages and Jira tickets.
“So when you go to this meeting, you now have everything that has provided the context,” he said to. “You don’t try to update, you can actually spend time on important strategy decisions.”
Atlassian estimates that this particular case of operate saves at least four hours per person a week. In particular, Harpercollins, in particular, used it for a “great effect” to notice.
Customers operate AI agents in various complexities, she said: sometimes they simply discharge their work, collect data or write notes from the issue; Another time they enter deeply into raw data and strategic road maps.
To explain that Atlassian has built a layer of chart on his data, which provides deeper intelligence about the connection of data. For example, enterprises can analyze their goals along with the structure of the team and in the course of projects. “It’s not just the ORG HR chart,” he said to.
“When you are wondering how people are now building a software development cycle, a huge part of this is the creation of road maps and prioritic strategies,” she said. “But this can be very dynamic, and taking into account all this data is difficult for people. The agents we see now become very popular among customers, they actually prepare these strategic road maps.”
To emphasize the importance of creating feedback with clients, noting that in the last three months Atlassian users have adapted 10,000 different versions of the company’s unchanging agents.
“This is a really great pool of feedback, which then helps us understand how they deposit these agents in work flows,” he said to. “I think that part of what is really exciting in this wave is that it is such a common design process with clients.”
Earning trust, building it from the very beginning
Kurien emphasized that trust is the cornerstone of any product and it should not differ in relation to artificial intelligence. Customers want to know what the agent does behind the scenes and have control over their activities. This requires strict review processes.
“New gaps have appeared with new waves,” she said. “We have built a solid process in which we identify a life cycle in which the agent enters and creates appropriate review processes for this phase.”
Emphasize the fact that this is more than raw technology; People have to cooperate, build complete solutions together and operate experience. The industry must invest in mighty data architecture and have the appropriate data context so that AI agents can make powerful decisions that we will ask for.
“The one where it becomes really exciting is that it is a superpower in your organization, when he is able to help in making better decisions, release better products, re -play goals, be more competitive as a company,” she said to.
She noticed that there were many waves of innovation over the years, but the one from AI is one of them. “I feel like in AI, it’s a fusing wave. It’s a moment after a while, right? And it’s just completely different than all other waves.”
Editor’s attention: As thanks to our readers, we opened early registration of birds for VB Transform 2026-Zaledwie 200 USD. Here, AI’s ambition meets with operational reality and you will want to be in peace. Book your place now.