Go to content

This website describes the services that BNP Paribas Asset Management delivers to AXA Group entities and joint ventures. 

AI solutions: Build In-House or Partner Externally?

As organisations grapple with AI adoption, one of the recurring questions I encounter — whether from established investment businesses or first-time founders — is whether to develop AI capabilities internally or turn to a service provider.

The answer depends less on hype and more on context: size, scale, and stage of the business.

Internal Development

Building AI solutions in-house can be powerful, but it demands:

  • Talent: Skilled data scientists, engineers and product leaders who understand both the technical and business dimensions.
  • Infrastructure: Reliable data pipelines, computing resources and compliance frameworks.
  • Time: AI models don’t just “work out of the box.” They require iteration, fine-tuning and ongoing monitoring.

For large-scale businesses, or those whose competitive edge depends on unique data or proprietary insights, investing internally can make sense. It creates defensibility and deep integration.

Service Providers

For early- to mid-stage organisations, external providers can accelerate adoption without the overhead of building a team from scratch. Service providers offer:

  • Speed to implementation: Leveraging ready-made frameworks and tools.
  • Specialised expertise: Access to niche knowledge without long hiring cycles.
  • Scalability: Ability to pilot use cases before committing significant internal resources.

However, outsourcing also has trade-offs: over-reliance can limit your ability to differentiate, and providers may optimise for generalised solutions rather than your specific needs.

A Balanced Approach

The most resilient path often lies in hybrid models:

  • Start with external providers to validate value.
  • Build internal capabilities as the business grows, focusing on areas where AI can create lasting competitive advantage.

BNP Paribas Asset Management’s solution for AXA clients and joint ventures 

Within the Operational Due Diligence (ODD) team responsible for AXA clients and joint ventures, we chose to follow a hybrid approach.  

With the team analysing between 400 and 500 funds per year, we sought to increase efficiency and productivity, allowing individuals to focus expertise on analysing important data. The solution, an online platform called CENTRL GPT, acts as a virtual co-pilot for analysts and managers, using AI as a learning language model. 

The tool makes it possible to assess the quality and relevance of the responses to the questionnaires sent by partner management companies in a few minutes, based on an evaluation methodology for information and key criteria defined internally. 

AI to enhance robust decision-making

AI is not about adopting the newest technology for its own sake. It’s about making disciplined decisions:

  • Does this add value beyond what we already do well?
  • Is this the right stage to invest heavily in internal development, or is an external partner more practical?
  • How does this align with our long-term strategy?

In the end, the question isn’t whether to adopt AI — it’s how to adopt it wisely.