Nexthink Experience ’26: executive briefing.

Executive overview

The question was people, not agents.

Pedro Bados, Nexthink’s CEO, skipped the usual “look what our AI can do” keynote. He asked instead what happens to people when AI can do almost anything, and his answer was that it still needs a human to give it purpose, judgment and accountability.

The keynote followed two made-up characters: Naveen, an IT leader whose job is turning from running technology into directing it, and Marie, a filmmaker whose team just needs the tools to work. Through them, Nexthink showed three upgrades you can use today and two new products due in 2027. The bigger story is where the platform is heading: from seeing and fixing problems to running the digital workplace, with a person signing off on anything that matters.

Available now · Nexthink Spark

Spark: from answering questions to getting things done.

Spark, the personal IT agent for every employee, launched a year ago and now serves millions of employees at hundreds of customers. Nexthink puts fully automated resolution at 70% today and says the new features are aimed at 80 to 90%.

• Spark Recommendations. Tells you what to fix to push resolution up. It spots knowledge gaps (the keynote example was 60 escalations from one rendering team asking how to share big files externally) and points to actions worth switching on or building.
• Adaptive learning (“Spark Dreams”). Overnight, Spark goes back over the day’s conversations and learns from them on its own, so answers get better without anyone curating them.
• Spark Cockpit. One view of rollout and adoption, resolution rates, and how support is shifting: fewer emails, portal tickets and phone calls, and fewer tickets touched by a human than a year ago.
• Spark app and open channels. A native app installs through the Nexthink Collector with a single checkbox. Spark also runs inside Microsoft Teams and Copilot, behind other chatbots through a hand-off API, and inside custom agents built in tools like Copilot Studio.
• Safety by design. Spark only runs actions an admin has approved from a catalog that spells out what each one does and what the employee will notice. You can add your own remote actions and workflows, and soon Tasks.

Available now · AI Drive

AI Drive: from counting AI use to proving it pays off.

AI Drive arrived last year to measure which AI tools people use. It now lets leaders set adoption targets and keep an eye on them, and Nexthink says most of its customers already run it.

• AI Adoption Goals. A business leader picks a template, names the tool and says what good looks like. In the demo, a creative director set a goal of 75% of her team using Adobe Firefly at least two days a week, with progress shown team by team so she could see who needed help.
• AI Governance. Shows compliance teams which AI tools employees are actually using, approved or not, so they can keep enterprise AI use in check.

Available now · Application Experience

The AHEAD framework.

AHEAD sits on top of Application Experience and scores each application on five pillars, so an app owner can judge it on more than uptime and speed (Nexthink docs).

1. AAdoptionAre the right people using it, and is usage growing?Surfaces Low uptake, onboarding delays
2. HHealthDoes it perform reliably?Surfaces Experience and performance issues
3. EEfficiencyHow well do people use it, and how much support does it generate?Surfaces Ticket burden, struggling groups
4. AAffordabilityAre we paying for licenses nobody uses?Surfaces License reclamation
5. DDelightHow do employees feel about it?Surfaces Sentiment and satisfaction gaps

AI insights sum up each pillar in plain language, and each one opens a Nexthink Assist conversation. In the keynote, AHEAD found new hires losing three days on average waiting for a license. Assist traced it to license assignment that nobody had automated, and the business owner handed it to IT as a task with a single sentence.

New product · 2027

Tasks: orchestrated outcomes.

Tasks is the layer that runs the work. You give an AI agent the outcome you want, and it carries the job end to end using Nexthink’s data, remote actions and Spark.

Most IT work is still done by hand because automation is expensive to build. Tasks brings that cost down and widens what an agent can take on.

• Built in plain language. You describe the workflow in conversation and edit it the same way. No code, no drag-and-drop flowcharts.
• Four kinds of step. Agentic, where the AI decides; deterministic, which runs the same code every time and is used for security and permissions; employee steps that go through Spark; and human steps for physical work or the big decisions.
• Oversight is built in. IT sees only the tasks assigned to them, every task keeps a full trace and audit log, and the employee, IT and Spark can all work in one conversation.
• Examples shown. Onboarding and transfers, device provisioning, application access, distribution lists, guest Wi-Fi, reclaiming SaaS licenses and service requests.

The keynote’s employee-transfer task.

1. Automatic

   A transfer in Workday kicks off the task.

2. Agentic

   An agent suggests hardware based on what the employee does and what people in similar roles use.

3. Employee

   Spark offers her two options. She picks a MacBook Pro instead of filling in a 34-field form.

4. Agentic

   An agent checks stock and places an order if needed.

5. Deterministic

   A step added with one sentence gives her the same apps as her new peers.

6. Human

   A technician sets up the device and hands it over, and Spark updates the task from the chat.

New product · 2027

Control: vulnerability readiness and response.

Control answers what Nexthink called the top ask from digital workplace leaders: every device patched, compliant and set up the way it should be, with changes landing daily instead of monthly.

Attackers are using AI to find and exploit vulnerabilities faster than a monthly Patch Tuesday cycle can keep up with. The press release adds that Control sets compliance and patch priority by persona and risk level.

• Posture. Watches compliance checks in real time (antivirus, firewall, local admin use and more), with different rules by region and persona. With Tasks, a failed check fixes itself and only reaches a person if the fix fails.
• Exposures. Takes the weekly list of hundreds or thousands of findings from security and ranks them by real exposure: who is affected, how often the vulnerable app is used and how.
• Rollouts. Pushes Microsoft Intune deployments out in rings, checks the experience at each stage and can check in with the employees affected, through Spark. In the demo, a secure-share update reached ring 4 in two days without a problem, while a VPN client upgrade stopped after 125 of 33,000 devices when Macs started dropping their connection.

The VPN case is the human-in-the-loop model in miniature. The agent offered two options, pause everything or carry on. After one clarifying question, the IT lead chose a third: carry on for the Windows devices that were fine, hold the Macs and open a task to find out why.

Why this matters

Impact by stakeholder.

The upgrades you can use today should lift self-service and the return on AI this quarter. Tasks and Control could change how service delivery and patching work in 2027.

Add it up and it comes back to the employee: fewer interruptions, quicker access to the tools they need and less time lost to IT process.