News and Product Updates

What a KDD Cup-Winning AI Architecture Means for Insurance Brokers

atif.khan@databraid.ai (Atif Khan) — Wed, 09 Sep 2026 14:39:37 GMT

Last month, we celebrated an incredible achievement: DataBraid CTO Atif Khan and his team, Null Hypothesis, won the KDD Cup 2026 Creative Track with their project, PiTrace. Their work explored how AI agents and humans can collaborate to solve complex analytical problems in a transparent and trustworthy way.

For us, however, the significance extends far beyond winning a prestigious global AI competition.

The ideas validated through that work represent many of the same principles that inspire our vision for Broker Intelligence.

The Real Problem Isn't Data

Insurance brokers already have data.

In fact, they often have too much of it.

Client information lives in the BMS. Policy details live in carrier portals.

Claims information exists somewhere else. Renewal opportunities arrive through emails. Important conversations happen in Teams, on the phone, and in documents spread across multiple systems.

The challenge facing brokers today is not a lack of information.

The challenge is transforming fragmented information into actionable intelligence.

Every day brokers spend valuable time searching for answers:

Which clients are most likely to churn?
Which accounts are underinsured?
Which renewals require attention first?
Which submissions have stalled?
Where are cross-sell opportunities hiding?
Which carrier relationships need attention?

The information needed to answer these questions usually exists somewhere.

The problem is that no person can constantly synthesize hundreds of data points across dozens of systems in real time.

From Systems of Record to Systems of Intelligence

Over the past thirty years, insurance technology has focused on digitizing workflows and storing information.

Broker Management Systems became systems of record.

Carrier portals became systems of transaction.

CRM solutions became systems of engagement.

But a new category is emerging:

Systems of Intelligence.

These systems do not merely store information. They continuously analyze, connect, and interpret information to help professionals make better decisions.

This is where Broker Intelligence begins.

At DataBraid, we believe the future broker's workspace will not be defined by navigating between disconnected applications. Instead, brokers will interact with an intelligence layer capable of understanding clients, policies, carrier relationships, agency operations, and market conditions as a connected whole.

The Lesson from PiTrace: Build the Analyst

One of the most interesting concepts behind PiTrace was a simple idea:

"Write the analyst, not the analysis."

Rather than building a rigid workflow for every possible problem, the team focused on creating reusable skills that could help an AI agent understand questions, gather evidence, evaluate options, validate conclusions, and ask for human guidance when appropriate.

Insurance presents a remarkably similar challenge.

No two brokerages operate exactly the same way.

No two client situations are identical.

No predefined workflow can anticipate every question a broker may ask.

That's why the future is not a collection of static reports and dashboards.

The future is an intelligent system capable of acting more like a skilled analyst.

One that can:

Understand context
Connect information across systems
Surface relevant insights
Explain its reasoning
Learn from feedback
Recognize uncertainty

The goal is not automation for its own sake.

The goal is better decisions.

Human Judgment Still Matters

One of the strongest themes that emerged from the KDD Cup-winning architecture was the role of human judgment.

The PiTrace platform was specifically designed to surface evidence, expose reasoning, identify uncertainty, and involve people when important decisions required interpretation or context.

We believe the insurance industry will follow a similar path.

Brokers are trusted advisors.

Their value does not come from data entry, portal navigation, or searching for documents.

Their value comes from judgment.

It comes from understanding clients.

It comes from interpreting risk.

It comes from building relationships.

The role of AI should be to amplify those strengths, not replace them.

The best brokerage of the future will combine the scale and speed of artificial intelligence with the experience and judgment of exceptional professionals.

The Vision for Broker Intelligence

When we talk about Broker Intelligence, we are not talking about a chatbot bolted onto a broker management system.

We are talking about a fundamentally different operating model.

Imagine an intelligence layer that can:

Continuously monitor activity across systems
Identify at-risk clients before they leave
Highlight account-rounding opportunities
Surface submission bottlenecks
Explain changes in retention performance
Recommend next best actions
Learn from brokerage-specific workflows and preferences
Provide transparent evidence for every recommendation

The result is not merely faster execution.

The result is a brokerage that becomes increasingly informed, proactive, and scalable.

Why This Matters Now

Artificial intelligence has reached an important inflection point.

The conversation is no longer about whether AI can generate text, summarize documents, or answer questions.

The real challenge is creating systems that can be trusted inside mission-critical workflows.

That requires more than powerful models.

It requires architecture.

It requires transparency.

It requires explainability.

It requires human oversight.

These were among the core ideas demonstrated through the KDD Cup-winning PiTrace platform, which emphasized evidence-based reasoning, reusable capabilities, human collaboration, and transparent decision-making.

Those same principles are helping shape how we think about Broker Intelligence at DataBraid.

Looking Ahead

Insurance is one of the most relationship-driven and knowledge-intensive industries in the world.

The next generation of broker technology will not be defined by larger databases or more dashboards.

It will be defined by intelligence.

Systems that can connect information, understand context, surface opportunities, and help brokers make better decisions.

Winning the KDD Cup was an exciting milestone.

More importantly, it reinforced a belief we hold strongly at DataBraid:

The future is not human or AI. The future is human and AI, working together.

And that's exactly the future we're building.

DataBraid CTO Atif Khan Wins KDD Cup 2026 | Human-AI Innovation

nick@databraid.io (Nick Romano) — Mon, 17 Aug 2026 21:14:18 GMT

We are incredibly proud to share that DataBraid CTO Atif Khan and his team, Null Hypothesis, have won the KDD Cup 2026 Creative Track, one of the world's most respected competitions in data science, machine learning, and artificial intelligence.

The achievement was announced at the ACM SIGKDD Conference on Knowledge Discovery and Data Mining in Seoul, South Korea, where Atif presented the team's winning submission, PiTrace.

Building the Analyst, Not the Analysis

The central idea behind PiTrace is both simple and profound.

Rather than building a system that produces answers from a black box, the team focused on building a platform where humans and AI agents work together to solve complex problems. As Atif explained during his presentation:

"We decided to write the analyst, not the analysis."

PiTrace was designed as a human-agent workbench where AI performs the heavy lifting of discovery, analysis, planning, validation, and problem solving, while humans remain actively involved in judgment, interpretation, and decision making. The result is a transparent, evidence-based approach to AI that makes reasoning inspectable, explainable, and trustworthy.

Why This Matters

As artificial intelligence continues to evolve, organizations face an important question:

How do you create AI systems that people can trust?

The Null Hypothesis team approached this challenge by building systems that expose evidence, validate conclusions, surface uncertainty, and invite human collaboration when judgment is required. Their work demonstrated that the future of AI is not simply about automation. It is about combining the strengths of intelligent systems with the expertise of human professionals.

In a recent reflection on the competition, Atif highlighted four practices that proved critical to the team's success:

Clear specifications that defined intent, constraints, and measures of correctness
Using multiple AI models in complementary roles
Treating implementations as replaceable while preserving evaluation frameworks and reusable skills
Focusing on when human judgment should guide, validate, or correct an AI system

These principles helped the team create a solution capable of tackling complex analytical challenges while remaining transparent and accountable.

A Proud Moment for DataBraid

While this achievement was earned by Atif and his KDD Cup team, the ideas validated through this work strongly align with the vision we are building at DataBraid.

Insurance brokers operate in one of the most information-intensive industries in the world. Every day they must navigate complex relationships between clients, carriers, policies, claims, communications, and regulatory requirements. The future belongs to organizations that can transform that information into actionable intelligence while keeping humans firmly in control of the decision-making process.

At DataBraid, we believe the most powerful systems are not those that replace professionals, but those that help professionals perform at their best. The same philosophy that helped earn recognition on one of the world's premier AI stages is reflected in our broader mission to build intelligent systems that augment expertise, improve decision making, and unlock new levels of productivity.

Congratulations

On behalf of the entire DataBraid team, congratulations to Atif Khan and the entire Null Hypothesis team on this remarkable accomplishment.

Winning the KDD Cup is a significant achievement in its own right.

More importantly, it represents the kind of innovation, rigor, and forward thinking that will help shape the next generation of intelligent systems.

We're excited to see where this journey leads next.

Congratulations, Atif. We're proud to have you leading our technology vision!

Building the On-Ramps to the API Superhighway: How Brokerages Can Prepare for the Future Today

nick@databraid.io (Nick Romano) — Thu, 09 Jul 2026 18:39:09 GMT

The Canadian insurance industry is on the verge of a massive technological leap. As discussed in earlier blogs, for years, the industry has discussed the "utopia" of Single Entry, Multiple Carrier Interface (SEMCI). Today, thanks to the incredible momentum driven by the Centre for Study of Insurance Operations (CSIO) and its carrier partners, the blueprint for that vision is actively coming together.

A Proposed Gateway to the Future

The progress coming out of CSIO's Innovation and Emerging Technology (INNOTECH) Advisory Committee over the last few years has laid critical foundations, including the delivery of 165 JSON API use cases in 2023 and the introduction of the industry's first API Security Standards .

Now, the industry is taking the next step. CSIO recently concluded a successful demonstration environment and is currently proposing an industry-wide API gateway—a "superhighway" that would allow broker management systems to connect once to CSIO, rather than building individual connections to every insurer .

We are incredibly optimistic about this initiative. If the industry unites and commits to moving forward with this model, it has the potential to fundamentally transform broker connectivity by Q2 2027 .

However, as with any massive infrastructure proposal, roads take time to approve and build. While Quote & Bind (new business) workflows are an excellent starting point for this API adoption , complex mid-term policy servicing naturally takes longer to standardize across every carrier's legacy systems. Brokerages cannot put their daily operations on hold while the industry aligns on these next steps.

DataBraid: Building the Car and the On-Ramps

If the industry agrees to pave this superhighway, DataBraid is building the unified broker workspace—the vehicle—that will travel on it.

We believe that brokerages should be able to embrace the API future without suffering through the operational drag of "portal hopping" today.

This is why DataBraid is engineered as a dual-pathway infrastructure .

Our unified workspace provides brokers with a single interface to initiate and manage servicing workflows. Behind the scenes, DataBraid translates the broker's intent into a "canonical" event log natively designed to consume CSIO's JSON and XML API standards.

Where a carrier has exposed a formal CSIO API, DataBraid routes the transaction directly. Where a carrier is still modernizing and relies on a legacy portal, DataBraid provides the on-ramp: we use secure, agentic and deterministic automation to execute the approved task inside the portal using the broker's own credentials.

Ready for Whatever Comes Next

The true magic of this approach is what happens as the industry evolves.

If the CSIO gateway is approved, or as more carriers successfully expose native APIs over the coming years, DataBraid can seamlessly hot-swap DataBraid portal connectors for the API connection behind the scenes. For the brokerage team, the workflow never changes. There is zero disruption, zero retraining, and zero downtime.

We are enthusiastic supporters of CSIO's vision and hope to see the industry rally behind it. In the meantime, DataBraid acts as the execution bridge, standardizing the broker's workflow today, fully supporting CSIO's security and governance standards, and accelerating the industry's journey toward whatever the API future holds.

DataBraid Raises $1.9M USD from Koru to Tackle Insurance’s Most Persistent Workflow Problem

nick@databraid.io (Nick Romano) — Tue, 09 Jun 2026 12:37:13 GMT

There’s a quiet inefficiency at the heart of the insurance brokerage industry—one that almost everyone acknowledges, but few have meaningfully solved.

Every day, brokers are forced to navigate a fragmented ecosystem of carrier portals just to complete routine work. Policy changes, endorsements, updates—tasks that should be simple become time-consuming, repetitive, and error-prone.

Today, DataBraid announced a $1.9 million USD pre-seed round to take on that problem directly.

__________________________

Read more about our funding and our unique approach to broker-first AI in our exclusive feature with BetaKit: HERE.

__________________________

The Operational Reality Brokers Face

For most brokerages, growth doesn’t just mean more customers—it means more operational load.

Each additional policy brings with it ongoing servicing requirements. And unlike sales, servicing doesn’t generate incremental revenue—it preserves it. Brokers typically earn around 15% commission on policies, so maintaining and growing a book is critical.

But the workflows behind that are far from efficient.

To complete even a basic servicing request, brokers often have to:

Log into multiple carrier portals
Re-enter the same information repeatedly
Navigate inconsistent interfaces and processes
Reconcile data between systems manually

It’s not just frustrating—it’s structurally expensive.

Over time, this creates real constraints:

Servicing teams become overwhelmed
Turnaround times increase
Margins compress as more headcount is required to keep up

The problem isn’t a lack of tools—it’s a lack of workflow cohesion across systems.

A Different Approach: Standardizing Work Above the System Layer

DataBraid is built around a simple idea:

What if brokers could complete servicing workflows in one place — and have that work propagate consistently across carriers?

Rather than replacing existing systems, DataBraid sits above them.

It creates a unified workspace where brokers can:

Enter and update information once
Execute workflows across multiple carriers
Avoid repeated portal logins and duplicate data entry

Under the hood, the platform standardizes workflows across carrier systems while respecting their:

Security models
Governance structures
Regulatory requirements

This matters. Insurance isn’t a blank-slate industry. Any meaningful solution has to work with existing infrastructure, not against it.

As Nick Romano, Co-founder and CEO of DataBraid, explains:

"Broker operations have become increasingly fragmented as servicing workflows spread across dozens of carrier portals. This round allow us to accelerate development of a broker-first layer that simplifies how work gets done without forcing carriers or brokers to change how they govern or secure their systems."

Building for Real-World Constraints

One of the reasons this problem has persisted is that it’s deceptively complex.

Carrier systems are deeply embedded. Data structures vary. Edge cases are everywhere. And above all, accuracy and compliance are non-negotiable.

DataBraid’s approach reflects that reality.

The platform combines:

Automation for repeatable workflows
Agent-assisted intelligence to guide actions
Explicit human approval to ensure accountability

This isn’t about fully replacing human workflows—it’s about removing the repetitive burden while preserving control and auditability.

In regulated environments, trust is not optional.

Why Now

The pressure on brokerage operations is increasing.

Books of business are growing, customer expectations are rising, and the operational burden of servicing is becoming harder to absorb. At the same time, the industry has reached a point where:

Integration standards are improving
Automation is more reliable
Brokers are more willing to adopt new workflow layers

The gap between how work is done and how it should be done is becoming impossible to ignore.

Industry Alignment

The round was led by Koru, the venture studio of the Ontario Teachers' Pension Plan (OTPP), reflecting a broader belief that this is not just a tooling issue, but a structural opportunity within insurance.

Importantly, the perspective from within the brokerage channel reinforces the urgency.

As Igal Mayer, CEO of Scoop Insurance, and an OTPP portfolio company put it:

"The inefficiency of juggling multiple carrier portals is one of the biggest operational drags on brokerage teams today. What’s needed is practical infrastructure that improves workflows without disrupting how the system fundamentally works."

That balance—improving efficiency without forcing systemic change—is what makes the problem both difficult and worth solving.

What Comes Next

With this funding, DataBraid plans to:

Expand its engineering team
Deepen integrations with carriers and broker systems
Advance early deployments across North America

For investors: We are actively building our network and would consider expanding this round for the right partner.

Join the Early Access ProgramIf you are tired of the portal-hopping grind, we are currently rolling out DataBraid to select brokerages. Join partners like McCAM Insurance and Begin in our Early Access Program to help shape the future of insurance workflows and start reclaiming your team's capacity today. Find out more and apply for the Early Access Program HERE!

The Bigger Picture

At its core, this isn’t just about portals or automation.

It’s about reclaiming time and focus.

When brokers spend less time navigating systems, they can spend more time:

Advising clients
Building relationships
Growing their business

The tools that enable that shift won’t just improve efficiency—they’ll reshape how brokerages operate.

And for an industry that has historically accepted fragmentation as a given, that shift is long overdue.

Read more about our funding and our unique approach to broker-first AI in our exclusive feature with BetaKit: HERE.

The full press release can be foundHERE.

_________

About DataBraid

DataBraid is a broker‑centric automation platform built to eliminate multi‑carrier portal work. Brokers enter information once in a unified DataBraid workspace and apply it seamlessly across servicing workflows and carrier interactions—without repeated logins, duplicate data entry, or constant portal switching.

By standardizing workflows above existing carrier systems, DataBraid improves speed and consistency while preserving carrier security, governance, and compliance requirements. Designed for regulated insurance environments, DataBraid emphasizes accountability, auditability, and trust at every step.

For more information, contact info@databraid.io.

About Koru Ventures

Koru is a venture studio backed by Ontario Teachers’ Pension Plan. Founded in 2019, Koru exists to drive value creation and grow the overall enterprise value of OTPP’s portfolio companies by partnering with them to identify, validate, build and oversee new disruptive business opportunities that leverage unfair advantages. Ultimately, this is in service of providing the pensions of the teachers of Ontario.

Since 2019, Koru has built and launched 11 new ventures across a variety of industries and geographies including child care, field safety, customer service innovation, financial services, e-commerce logistics, insurance, elder care and climate tech/energy transition.

The Utopia of SEMCI: Reimagining Brokerage Workflows

nick@databraid.io (Nick Romano) — Mon, 01 Jun 2026 19:12:17 GMT

The modern insurance brokerage runs on a paradox: while distribution has become more sophisticated, the core workflows powering it remain deeply fragmented.

At the center of this fragmentation is a simple but costly reality—brokers spend a disproportionate amount of time navigating carrier systems instead of serving clients. The idea of SEMCI (Single Entry, Multiple Carrier Interface) represents a kind of operational utopia for the industry: enter risk data once, and transact seamlessly across all participating insurers.

It’s a vision that has existed for years. But achieving has remained elusive for decades.

The Problem: Fragmentation as the Default

Insurance brokers sit in a unique position. They are advisors, salespeople, and service agents—but critically, they are also the connective tissue between customers and carriers.

Yet each carrier operates within its own ecosystem:

Separate portals
Unique underwriting rules
Different data requirements
Inconsistent APIs or integrations

As a result, even basic workflows—quoting, binding, endorsements—require repeated data entry across multiple systems.

This is especially painful in policy servicing, where the work is less visible than sales but just as critical. Address changes, coverage updates, vehicle swaps, driver adjustments—each request often triggers portal hopping, follow-ups, and manual reconciliation.

The inefficiency compounds:

Data is rekeyed multiple times
Context is lost between systems
Errors creep in
Turnaround times increase

In an industry where responsiveness and accuracy define client trust, this creates a structural disadvantage.

The Economic Reality: Thin Margins, High Friction

Most brokerages operate on a relatively simple revenue model:
they earn commissions from carriers—typically around 10–20%, with ~15% as a common benchmark—on the policies they place and maintain.

This means:

Revenue scales with policy count and retention
Profitability depends heavily on operational efficiency

Every minute spent inside a carrier portal is effectively a cost.

Policy servicing, in particular, becomes a silent margin killer. Unlike new business, it doesn’t generate incremental revenue, but it:

Consumes staff time
Requires experienced resources
Cannot be avoided (it’s core to retention)

As books of business grow, servicing load grows with it. Without efficiency gains, brokerages face a ceiling where:

More clients require more staff
Margins flatten or compress
Growth becomes operationally expensive

In other words, the current model doesn’t scale cleanly.

Portal Hopping: Death by a Thousand Clicks

“Portal hopping” is not just an inconvenience—it’s a systemic drag.

A single servicing request might look like this:

Open the BMS to understand the policy
Log into Carrier A’s portal
Re-enter data
Submit the change
Repeat for additional carriers or confirmations
Update the BMS manually

Multiply that across:

Dozens of carriers
Hundreds of daily transactions
Thousands of policies

The result is hours of low-value work disguised as necessary process.

And critically, this work is not differentiated. It doesn’t improve advice, deepen relationships, or drive revenue. It is purely transactional overhead.

The SEMCI Vision: Enter Once, Transact Everywhere

SEMCI proposes a radically simpler model:

Capture structured risk and policy data once, then distribute it programmatically across carriers.

At its best, SEMCI would:

Eliminate redundant data entry
Standardize submission workflows
Enable real-time or near-real-time interactions with carriers
Keep the Broker Management System (BMS) as the operational source of truth

In this model, the broker’s workflow becomes:

Enter or update client data once
Trigger transactions across one or more carriers
Receive structured responses back into the system

No portal switching. No duplicate effort.

Justflow.

Why It’s Hard: The Reality Behind the Utopia

If SEMCI is so obviously valuable, why hasn’t it been fully realized?

The challenges are both technical and institutional:

1. Data Standardization

Insurance data is notoriously inconsistent. Even something as basic as a vehicle or driver record can vary in structure and terminology across carriers.

Efforts like CSIO standards have made progress, but gaps remain—especially in edge cases, endorsements, and underwriting nuances.

2. Carrier Incentives

Carriers invest heavily in their own portals, which:

Provide control over user experience
Enforce underwriting discipline
Capture broker behavior data

Opening up fully to standardized, external workflows can feel like a loss of control.

3. Legacy Systems

Many carrier back-end systems were not designed for real-time interoperability. Retrofitting them to support seamless, structured transactions is non-trivial.

4. Edge Case Complexity

Insurance is full of exceptions. Even if 80% of transactions can be standardized, the remaining 20%—complex risks, manual underwriting, endorsements—can break automation flows.

5. Trust and Accuracy

For brokers, accuracy is everything. Any SEMCI system must match or exceed the reliability of current workflows, or it introduces risk instead of eliminating it.

The Payoff: What Success Looks Like

If SEMCI were fully realized, the impact on the brokerage landscape would be profound.

1. Margin Expansion

Reducing servicing time directly improves operating leverage. Brokerages could:

Handle larger books with the same staff
Improve profitability per policy
Reinvest in growth or client experience

2. Refocused Talent

Instead of spending time on data entry and navigation, brokers could:

Advise clients more deeply
Cross-sell and upsell effectively
Strengthen retention

3. Faster Service, Better Experience

Clients benefit from:

Faster turnaround times
Fewer errors
More responsive service

4. Scalable Growth

SEMCI unlocks a new growth curve where:

Adding policies does not linearly increase operational burden
Consolidation becomes easier
Brokerages can compete on strategy, not just execution

From Ideal to Inevitable

SEMCI may sound like a utopian concept, but it reflects a broader truth:

Every mature industry eventually optimizes around data reuse and workflow efficiency.

Insurance is simply earlier in that journey.

The pressure is mounting—from broker economics, customer expectations, and technological progress. While full SEMCI may not arrive overnight, incremental progress—better integrations, more structured data, improved interoperability—is already reshaping the landscape.

The brokers who benefit most won’t just be those who wait for the future.

They’ll be the ones who recognize that the cost of inefficiency is compounding—and that solving it is no longer optional.