We launched spektr 2.0 last week – all about agentic AI, a modular approach, and fundamentally being able to build whatever compliance workflows you need with a singular platform. But we left out the biggest spektr 2.0 product update, saving it for its own spotlight: introducing our brand new AI Agent Builder.

As we spent more time with clients, we noticed a pattern. They weren’t just asking for automation. They were asking for a way to shape intelligence – to tell the system exactly how they wanted decisions made, what data to use, and when to involve human judgment. They wanted to scale expertise, not just rules. That’s what led us to the AI Agent Builder.

Bringing custom AI agents to life

Until now, using AI for complex, context-heavy compliance decisions was seen as risky. Most teams either relied on rigid, rule-based workflows that couldn’t adapt quickly, or avoided AI altogether because it wasn’t clear enough how it reached its conclusions. For regulated industries, that simply isn’t an option.

The Agent Builder changes that.

It’s an intuitive, workflow canvas where you can design AI agents in minutes. You start with a builder prompt – a clear instruction describing the agent’s goal. Maybe it’s monitoring clients’ websites for industry changes, verifying customer identity, scanning corporate registries, or detecting emerging fraud patterns.

From there, you wire in:

• Data sources (your own, or from external search)
• Logic prompts that outline decision rules
• Human-in-the-loop review points for riskier cases

• AI routing steps that branch based on outcomes

Every agent is transparent by default. Before it runs, you see its checklist of actions. After it runs, you see the reasoning behind every conclusion, the data it used, and the policy references it followed. If you disagree, you can override it — and the agent learns from your correction.

It’s AI you can direct, audit, and trust. As our CPO, Jeremy Joly, mentions:

“Successful adoption comes from choice. With the Agent Builder, you decide what’s automated, what stays in human hands, and how that balance shifts over time based on your risk appetite, industry, or jurisdiction. We don’t force AI into your workflows…we make it possible to put it exactly where it adds value, and keep full control everywhere else.”

How did it all begin?

When we built the first version of spektr, our goal was to give compliance teams clarity, automation, and modularity. You could configure processes without code, connect external data, and be audit-ready from day one – because each business is different, including interpretation of risk, making compliance and risk highly case-specific.

But across the industry, one challenge persisted: compliance decisions aren’t always binary. They require context, interpretation, and a deep understanding of the business. Hardcoding every possibility was never practical, and every regulation change risked triggering costly redevelopment. Integrating new data sources often meant delays measured in weeks, not hours.

What we heard from clients was consistent with this broader reality: they wanted AI that could follow their policies to the letter, adapt instantly when rules changed, and work with their own data, thresholds, and oversight. They wanted visibility – not just an answer, but the reasoning behind it, whether it aligned with policy, and the ability to adjust it in real time.

It’s exactly this industry gap, validated by our clients’ feedback, that inspired the creation of the Agent Builder.

Rethinking the foundation: what the Agent Builder unlocks

With spektr 2.0, we rebuilt the foundation of compliance to be modular. With the Agent Builder, that modularity now extends into AI itself. Here are just a handful examples of how you could use the Agent Builder in the wild:

• Monitoring clients for industry changes based on online signals
• Building an onboarding risk agent that checks dozens of risk factors, explains each one, and routes edge cases to an analyst
• Creating a SAR-filing agent that drafts suspicion narratives, fills in regulatory forms, and sends them for review
• Detecting and escalating transactions to/from FATF high-risk countries

And no, these aren’t pre-built templates. They’re agents you define with your data, your policies, and your thresholds. And because they give full reasoning, they provide the transparency regulators demand.

AI you can certify

The Agent Builder isn’t just a technical upgrade, it sits atop a governance foundation already certified to ISO 27001 (information security) and ISO 42001 (AI management systems). spektr operates with an integrated management system covering both standards, and we’re proud to be among the first ten companies globally to achieve ISO 42001 certification, thanks to Mastermind. That means every agent you design inherits the same third-party-audited controls for security, transparency, and responsible AI use.

Where we’re going

The AI Agent Builder isn’t just another feature. It’s the next step in everything we’ve been building since day one: compliance infrastructure that gives teams superpowers.

In a world where regulatory expectations evolve weekly and risks emerge overnight, the ability to design your own intelligence is a competitive advantage. With the Agent Builder, you decide how far automation goes, where human oversight steps in, and how quickly your approach adapts.

As our CEO, Mikkel Skarnager, puts it:

“The Agent Builder gives compliance teams the power to design intelligence as easily as they design processes. It’s not just about replacing manual work — it’s about creating AI that works exactly the way your business does, with transparency and control built in from day one.”

With the AI Agent Builder, you’re not just automating workflows. You’re building a team of tireless, transparent, and fully compliant AI analysts – ones that work exactly the way you want.