AI you can understand, control and defend

Artificial's view on AI in insurance is simply that it should be used to enhance expertise, rather than obscure or replace it.

We know that AI is becoming a routine part of how specialty and commercial insurance is done. Our partners use it to interpret large volumes of information, remove repetitive tasks, and help brokers and underwriters move through complex workflows faster.

However, we believe that speed and efficiency should not be the only considerations. Insurance decisions are complex and need to be thoroughly auditable, justifiable, and comprehensible. When a decision is questioned, the people responsible must be able to explain how it was reached.

Underwriting and broking are not simply a series of data points from which a machine can calculate the 'right' answer. They involve experience, context, commercial judgement, negotiation and an understanding of risk that has been developed over many years.

That's why our approach to AI starts with a simple principle: your rules should govern the system, not the other way round.

White-box, not black-box AI

There is a fundamental difference between using AI to support a decision and handing that decision over to an opaque, black-box model. At Artificial, we believe AI in insurance should be transparent by design.

We designed our AI to work within the rules, controls and underwriting strategy defined by our partners. It can still be used to help interpret information, identify what matters and remove repetitive work, but the professional's logic and expertise remain firmly at the centre of the process. We call it white-box decision-making.

By doing this, automation is not used for automation's sake, but rather used in a way that keeps teams more effective whilst maintaining clear oversight of how and why decisions are made.

Why we created AgLabs

AgLabs is the agentic arm of Artificial. It deploys digital workforces into insurance organisations today and is building the foundations for agent-to-agent (A2A) coordination across the market.

We created a distinct AgLabs offering because agentic work has a different shape from a conventional platform deployment. It often starts with a focused, client-specific workflow. The aim is to put useful outputs in clients’ hands within days, rather than weeks or months, so they can assess the results in their own workflows and build from there.

This allows a client to deploy an AgLabs agent where there is an immediate opportunity, whether or not it needs the full Artificial platform. It also gives us a clear environment in which to develop emerging workflows with forward-looking clients, learn from controlled production use and expand only when the technology and operating model are ready.

Artificial provides proven digital placement and underwriting infrastructure. AgLabs provides agentic services and digital workforce deployment. They are part of the same company, share the same insurance expertise, and can work independently or together.

Building the foundations for a 24/7 agentic insurance ecosystem 

The team at AgLabs is investigating how AI agents can handle more of the administrative and technical work surrounding insurance transactions, and how agent-to-agent communication (A2A) can accelerate coordination between market participants and create a digital interaction layer across the insurance market.

In an A2A market, an agent representing one participant can coordinate directly with an authorised agent representing another. A broker agent and an underwriter agent are one obvious example, but the same model can extend across carriers, MGAs, coverholders, reinsurers and other market participants.

Individual AI can make a person faster. A2A can make the interaction layer of the whole market faster.

The market is quickly catching up to the possibilities of this technology. During an Instech event this year, 49% of poll respondents said they believed AI agents could handle more than half of London Market interactions by 2030. A further 33% expected agents to handle between 25% and 50% of interactions.

It's clear that organisations must, therefore, have full oversight of what agents can achieve, what information they can access, which rules govern their behaviour, and when decisions must be referred to a person. Agentic technology should not create an even bigger black box. As its roles grows, transparency, traceability and human oversight become even more crucial.

From research to real-world results

AgLabs is already live and integrated within the workflows of some of our insurance partners, helping teams to gather information, process documents and prepare work for human review.

Early results include:

• 50 hours saved per week across a single line of business using custom agents in production
• 6 custom agents live, built and deployed within weeks
• 300 emails and 2,500 attachments processed in just one month

These results demonstrate the practical potential of agentic AI in insurance today. The aim of AgLabs is not to automate judgement, but to remove the friction around it. This gives market participants more time to focus on risk, relationships and decision-making.

Start a conversation

If you're exploring how AI agents could support your underwriting or broking workflows while maintaining a white box, transparent approach, get in touch with our team.

Talk to us