Being an MGA CTO: Lessons from the Front Line

Introduction: Accidental CTOs and Entrepreneurial MGAs

When people ask me what life was like as a CTO of an MGA, my honest answer is: it was never really a CTO role in the way Silicon Valley would define it. It was closer to being a business technologist, part CIO, part product lead, part translator between underwriters and engineers, and occasionally the referee.

I co-founded Azur Underwriting in January 2016 with Graham Elliott. We started in a coworking space near Finsbury Square, two founders with a blank sheet of paper, and the sort of optimism that only exists before you’ve tried to automate insurance.

Our very first technology decision was almost comically small: we needed to buy laptops.

The Laptop That Set the Tone

We walked into PC World and faced a binary choice: Windows or Apple.

Graham pushed hard for Macs. His reasoning was simple but profound: if we live in the browser and live in the cloud, we force transformation. His experience running an MGA had been shaped by legacy systems, poor user experience, and technology that dictated behaviour rather than enabled it.

I didn’t mind either way. I’d come from digital marketing and web businesses, where browsers were the operating system. Files, desktops, and spreadsheets weren’t central to how work got done.

This single decision quietly defined our technology thinking:

• Cloud-first by default
• Browser-based workflows
• A bias towards modern UX

It also made us deeply unpopular with some colleagues.

The Spreadsheet Wars (Which Excel Won)

We tried to ban spreadsheets.

That didn’t go well.

Excel is arguably one of the most valuable pieces of software ever written, but it’s also a terrible automation strategy. We were trying to run billing off an 80,000-row spreadsheet on Excel for Mac, which broke frequently and terrified everyone involved.

Reality won. Finance got Windows machines. Excel stayed.

The real lesson wasn’t that spreadsheets are bad.

It was this:

If you want to automate middle- and back-office processes, Excel cannot be your system of record.

It’s brilliant at analysis and modelling, but dreadful at scale, auditability, and integration, all things MGAs eventually care deeply about.

Learning Insurance from the Inside

I had no background in insurance. I understood regulated businesses from investment banking, but underwriting was new territory.

What struck me early on was how much human effort existed around systems that were otherwise reliable and cheap. Our core system, from a great little company called Anodas‚ was robust, auditable, low-cost, and still in use today.

But it mostly handled numbers and words.

People did the automation.

Teams manually assembled policy packs. PDFs were emailed. Brokers couldn’t access systems directly. This wasn’t broken‚ it was just how insurance worked.

Even today, much of the industry still relies on armies of people loading data: whether in-house, offshore, or via BPO providers. AI is now replacing some of that work, but in many cases it’s just swapping a human ‘robot’ for an agentic one.

Underwriters vs Technologists: The Necessary Tension

Every MGA CTO quickly discovers an uncomfortable truth:

In underwriting businesses, underwriters are the power centre.

Nor should that surprise anyone.

The challenge is building mutual respect between underwriting and technology. Agile ceremonies like ‘show and tell’ often turned into ‘shut up and listen’ as underwriters focused on edge cases and correctness while engineers tried to think in systems and abstractions.

Both sides were right.

Underwriters are paid to be precise. Engineers are paid to generalise. Respect had to be earned albeit slowly.

Buy vs Build: The MGA Reality

As a small MGA, the classic enterprise question ‘build or buy’ takes on brutal clarity.

We spoke to Duck Creek. The conversation ended quickly. The entry ticket was measured in millions we didn’t have.

Building our own core system made no sense either. Insurance systems are mostly:

• Systems of record
• Maths
• Versioning
• Document generation

Not trivial, but not hedge-fund-level complexity either.

The real problem wasn’t functionality. It was architecture, UX, and economics.

Why Salesforce (and Why So Early)

In 2016, cloud platforms were still maturing. AWS was accelerating. Digital transformation was the buzzword of the day, and arguably still is. COVID-driven remote working hadn’t yet forced the issue.

I had used Salesforce before, and what caught our attention was the emergence of vertical platforms and particularly a company called Vlocity that was a Salesforce app with a full insurance data model and basic processes.

The alternative was painful:

• Policy admin system
• Salesforce for CRM
• Finance system
• Middleware to glue it all together

For a £50m GWP MGA, middleware alone could destroy your margin.

Salesforce offered a provocative idea: one platform, one data model, one place to build.

It was early. It was imperfect. But strategically, it was the right bet.

Ironically, what many MGAs are trying to do today with modern insurance clouds is exactly what we were aiming for back then. It just took the ecosystem a decade to catch up.

Winning Credibility the Hard Way

Our first major technology project wasn’t even for ourselves.

It was for Ajit Jain at Berkshire Hathaway, arguably the most respected figure in insurance. Graham sold him on a vision centred on user experience and clarity, born out of frustration with complexity and small print.

That project didn’t just deliver value. It gave us credibility.

In fact, we had to persuade the most famous man in insurance before some of our own colleagues were convinced we knew what we were doing.

Good Enough Tech

One of the most important lessons I learned as an MGA CTO is this:

Perfect technology will kill you. Good-enough technology might save you.

We couldn’t afford to re-platform an entire legacy book. It was too complex and too expensive.

But we could:

• Launch new products on modern technology
• Prove economic value
• Incrementally shift behaviour

That approach worked. We built our fully digital home product, SmartHome. We demonstrated what was possible.

Eventually, Azur Underwriting was acquired by Aviva ‚ and now as tech business we still have a great relationship with them.

Borrowing Strength from What You Know

One of our smartest early moves had nothing to do with underwriting systems.

We launched BrokerIQ.

I had previously co-founded BrightTALK, a webinar platform used heavily in financial services. The parallels were obvious: fund managers talking to financial advisers weren’t that different from underwriters talking to brokers.

Broker IQ:

• Solved a real broker problem (CII accreditation)
• Demonstrated digital maturity
• Positioned us as a modern MGA

It cost around £100k, a serious investment at the time, but it paid off in perception, engagement, and credibility.

Reflections for Today’s MGA CTOs

Looking back and now through the lens of Azur Technology and modern insurance platforms, a few truths stand out:

• MGAs must be entrepreneurial by design
• Technology is a partnership, not a power grab
• UX matters more than most people appreciate
• Cost models matter more than architecture diagrams
• Engineers and underwriters are both perfectionists, just in different ways

Most importantly, success comes from focus. Solve the important problems first. Accept that some things only need to be good enough.

If you get that balance right, you don’t just build systems, you build trust. And in insurance, that’s still the hardest thing of all.

Takeaways for Today’s MGA Tech Leaders in an AI World

The technology landscape MGAs now operate in is very different from 2016 but many of the underlying lessons remain the same. AI hasn’t removed complexity; it has shifted where it sits.

Here are the takeaways I’d offer to anyone leading technology in an MGA today:

1. AI Is a Force Multiplier, Not a Strategy

AI will not fix broken processes. It will simply automate them faster.

Before deploying AI for submission ingestion, quote loading, or claims triage, be brutally honest:

• Do you understand the process end to end?
• Is it worth automating?
• Are you genuinely improving outcomes for underwriters, brokers, or customers?

The MGAs winning with AI are the ones using it to remove friction, not just cost.

2. Great Tech Is Now Table Stakes for an Attractive Exit

The idea of ‘good enough tech’ worked when MGAs were judged primarily on underwriting performance and distribution.

That world has changed.

Today, for MGAs looking to sell at an attractive multiple, great technology is table stakes.

AI lowers the cost of experimentation, but it raises expectations. Buyers now assume:

• Modern, cloud-native platforms
• Clean, accessible data
• Embedded automation and AI-ready workflows

You still cannot afford to:

• Replatform everything
• Rewrite the past
• Perfect every edge case

But you can, and must, demonstrate that new products, new flows, and new propositions are built on scalable, future-proof technology.

Legacy books can age gracefully. Growth books must look like the future.

That distinction increasingly drives valuation.

3. Data Model First, Models Second

Large language models are impressive, but without a clean data model they are theatre.

The hard work is still:

• Defining canonical data
• Owning your system of record
• Knowing where truth lives

AI sits on top of that foundation. It cannot replace it.

4. Underwriters Are Still the Power Centre

AI does not change the fact that underwriting judgment is the product.

The role of technology leadership is not to overrule underwriters, but to:

• Give them leverage
• Reduce cognitive load
• Make judgment scalable

Respect beats disruption every time.

5. UX Matters More Than Model Accuracy

A mediocre model embedded in a great workflow will beat a brilliant model no one trusts.

In insurance, adoption always wins over elegance.

If users don’t understand why an AI made a recommendation, they won’t use and they’ll find a spreadsheet instead.

6. Platforms Beat Point Solutions

The AI tooling market is exploding, but MGAs should be wary of assembling brittle stacks.

Each new vendor:

• Adds cost
• Adds risk
• Adds integration debt

Platforms that combine workflow, data, security, and AI will outlast clever tools that only solve one problem.

7. Tech Leadership Is Still Translation

The most valuable skill for an MGA tech leader hasn’t changed:

Translating underwriting intent into executable systems.

AI doesn’t remove the need for this role, it amplifies it.

Those who can speak underwriting, engineering, compliance, and economics at the same time will define the next generation of MGAs.

Written from experience, shaped by Azur Underwriting, and informed by a career spanning regulated industries, SaaS, and insurance technology.