Case Studies

Building a vertical AI venture platform for regulated professions

Turning one recurring market problem into a platform for creating specialised AI businesses

Sector

Regulated professional services

Client

Challenge

Regulated professions need AI, but generic tools often fail on privacy, sector context and professional workflows

Solution

Shared private AI platform used to create specialised applications for individual professions and industries

/

/

Building a vertical AI venture platform for regulated professions

/

/

Building a vertical AI venture platform for regulated professions

01

The opportunity

The same problem kept appearing in different professions.

Across legal services, healthcare, financial services and other regulated sectors, we saw a similar pattern.

Demand for AI was growing.

So was the pressure on professionals.

Teams were dealing with:

  • Increasing administrative work

  • Talent shortages

  • Growing document volumes

  • Higher client expectations

  • More regulation

  • More fragmented organisational knowledge

AI could help.

But generic AI products were often a poor fit.

Sensitive information could not simply be sent into consumer tools.

Professional workflows depended on specialist language and judgement.

AI output needed to be reviewable.

Organisations needed to know where information came from and who remained responsible for the final decision.

The problem was larger than one industry.

That made it interesting.

02

The venture thesis

Do not build the same AI foundation fifty times.

A legal AI product and a healthcare AI product may look very different to the user.

Underneath, they need many of the same things.

Secure document processing.

Knowledge retrieval.

Permissions.

Model management.

Workflow automation.

Human approval.

Logging.

Traceability.

Private infrastructure.

Instead of rebuilding those foundations for every industry, we created a shared platform that could support multiple specialised products.

The technology stays reusable.

The domain knowledge changes.

03

What we built

A platform for vertical AI.

The platform combines a shared technical foundation with sector-specific applications.

At the core sit capabilities for:

  • private AI environments

  • document ingestion

  • retrieval-augmented generation

  • semantic search

  • model orchestration

  • workflow automation

  • role-based access

  • activity logging

  • human approval

  • system integrations

Individual products can then add the knowledge, workflows and interfaces required by a specific profession.

04

The model

Shared technology. Specialised products.

The venture model has three layers.

1. Common infrastructure

A reusable AI platform provides the technical foundation.

2. Domain expertise

Professionals from each sector help define how the work actually happens.

3. Vertical products

Specific applications are built around the needs, terminology and workflows of that market.

This creates a repeatable path from sector insight to working software.

05

Why regulated markets?

The need for automation is high. So is the cost of getting it wrong.

Regulated professions often combine several characteristics that make them especially interesting for specialised AI.

They are knowledge-intensive.

A large part of the work happens through documents and information.

Specialist talent is expensive and difficult to scale.

Administrative tasks consume significant professional time.

Historical knowledge is valuable.

Privacy matters.

Professional accountability cannot simply be delegated to software.

That creates a strong case for AI, but also a high bar for how it needs to be built.

06

From generic AI to vertical intelligence

A general-purpose model starts with broad knowledge.

A vertical AI product needs more.

It needs to understand:

  • The profession

  • The terminology

  • The documents

  • The workflows

  • The permissions

  • The regulations

  • The organisation's own knowledge

The platform therefore combines general AI capabilities with private organisational data and sector-specific workflows.

The aim is not to create a model that knows everything.

It is to create a system that is useful inside a particular profession.

07

Domain experts become part of the product

Sector knowledge cannot be added at the end.

A common mistake in vertical AI is to build the technology first and ask professionals to validate it afterwards.

We took the opposite approach.

Domain specialists can participate in:

  • Opportunity selection

  • Workflow mapping

  • Product design

  • Knowledge structuring

  • Output evaluation

  • Testing

  • Human review

  • Market development

Their expertise shapes what gets built.

The current venture model explicitly brings engineers and experienced industry professionals together to create and validate sector-specific workflows.

08

Human judgement stays in the system

Automate the work around the professional, not the responsibility of the professional.

The platform is designed around human review.

AI can help:

  • Retrieve information

  • Structure dossiers

  • Prepare documents

  • Analyse files

  • Draft communication

  • Identify relevant information

  • Coordinate workflows

Professionals remain responsible for sensitive decisions and final approval.

That principle is reflected in the platform's governance approach, which states that AI should support human decision-making and that users retain control over AI outcomes.

09

One platform, multiple sectors

The shared platform can be applied to different regulated environments.

Current areas include:

Legal and notarial services

Private knowledge, document preparation, research and professional workflows.

Public sector

Secure information access and administrative workflows around public data and services.

Financial services

Sensitive document processing, knowledge support and regulated operational workflows.

Healthcare

Private AI around clinical and operational information with human oversight.

Insurance

AI-assisted workflows around policies, claims, documentation and customer service.

Each market gets its own product experience.

The technical foundation remains shared.

10

The first vertical proved the model

Start with one profession. Learn. Reuse what works.

The first major vertical focused on civil-law notary offices.

That product was built around confidential legal work, combining:

  • Private AI

  • Professional knowledge

  • Document workflows

  • Source-based answers

  • Human review

Working closely with professionals helped validate both the technology and the venture model.

The lessons from one vertical could then improve the shared platform used by the next.

This is where the model begins to compound.

11

The compounding platform model

Each new vertical creates more than a new product.

It can also improve the underlying platform.

A capability built for one sector may later become reusable infrastructure.

For example:

  • Document ingestion

  • Knowledge retrieval

  • Permissions

  • Workflow orchestration

  • Audit logging

  • Human approval

  • Evaluation

Instead of each company starting from zero, new ventures begin with an increasingly mature technical base.

12

From project economics to platform economics

A different model from custom software delivery.

Traditional custom development has a simple relationship:

  • New customer

  • → New project

  • → New build

The vertical platform changes that model.

  • Shared platform

  • → Sector-specific configuration

  • → Reusable workflows

  • → New product

Engineering investment can therefore benefit more than one product.

That creates the possibility for faster launches and more reusable intellectual property over time.

13

How The Forge contributed

Build the company around the technology.

The work went beyond software development.

The Forge model brought together:

Venture thesis

Identify a recurring problem large enough to support multiple products.

Market selection

Find professions where the problem is urgent and domain knowledge creates defensibility.

Product strategy

Define the common platform and the vertical product layer.

Brand and positioning

Create a proposition around private AI for regulated work.

Engineering

Build the shared technology and initial applications.

Domain partnerships

Bring professional experts and early customers into product development.

Go-to-market

Create the commercial model for launching sector-specific solutions.

Venture development

Build the organisational structure needed to grow beyond an individual project.

14

The result

From a single opportunity to a repeatable venture model.

The result is a platform designed to support multiple AI businesses serving regulated markets.

One reusable technical core

Common AI, knowledge, workflow and governance capabilities do not need to be recreated for every sector.

Stronger sector relevance

Each product can be designed around the actual work of a specific profession.

Domain-led development

Industry professionals help shape and validate the workflows.

Private by design

Dedicated environments, access controls and European data handling are built into the platform model.

Human control

Professional judgement remains part of sensitive workflows.

Faster path to new verticals

New sector products can start with an existing technical foundation rather than a blank page.

15

Why this matters

The next large AI companies may be built vertically.

General AI platforms are getting more capable.

That does not remove the need for specialised products.

In many professional markets, it increases it.

The model provides intelligence.

The vertical company provides:

  • Context

  • Workflow

  • Trust

  • Integration

  • Governance

  • Distribution

  • Domain expertise

That is where much of the product value sits.

This venture was built around that idea.

AI services used

The Forge

Turning a repeatable market insight into a platform, products and new venture model.

AI Strategy

Defining where a shared AI platform could create value across multiple industries.

Software Engineering

Building the platform and specialised vertical applications.

Data Infrastructure

Creating private knowledge, retrieval and data foundations.

AI Adoption

Embedding human control, data governance and responsible AI into the platform.

Avatar-1
Avatar-2
Avatar-4
Avatar-4
Avatar-3

Know a market that deserves its own AI company?

The strongest vertical AI opportunities often start with deep knowledge of a profession and a problem outsiders overlook. We bring the product, technology and venture capability around it.