Creating Value in Software Development Relationships in the Age of GenAI

Large, complex software development programs have always struggled with the same problem:

we are exceptionally bad at sizing, estimating, and governing them.

Despite decades of process maturity, tooling advances, global delivery improvements and user experience models, large programs still routinely miss cost, timeline and value expectations. The industry response has been remarkably consistent: more controls, more SLAs, more reporting—and, ultimately, more finger-pointing.

Now GenAI enters the picture.

Will GenAI Lead to Better Outcomes?

Maybe, if we do it right.

But in the near term, GenAI may actually increase missed expectations, not reduce them.

Why? Because technology has never been the primary driver of success or failure in large software programs. The determining factors remain:

• The quality of the client–provider relationship

• Trust and transparency

• Delivery and domain expertise

• The ability to adapt when assumptions break down

GenAI can accelerate code, automate tasks and improve quality throughput. What it cannot do—at least not yet—is resolve misaligned incentives, poorly structured contracts, or adversarial delivery models.

The Contracting Model Is the Real Bottleneck

Most managed services agreements still rely on a familiar structure:

• FTE-based pricing

• A narrow set of SLAs

• Vague or aspirational productivity and benchmarking commitments

This model was already suboptimal before GenAI. In an AI-enabled delivery environment, it becomes actively counterproductive.

Why? Because GenAI introduces uncertainty:

• Productivity gains are uneven and volatile

• Benefits vary by domain, team maturity, and architecture

• Measurement is difficult, especially early on

And uncertainty is exactly what traditional contracts are designed to avoid.

Why Risk/Gain Share Hasn’t Taken Hold (Yet)

Risk/gain-sharing models are frequently discussed—and rarely implemented well. The reasons are well known:

• Measurement complexity

• Disagreements over risk allocation

• Governance and data transparency challenges

• Fear of paying for “missed” objectives

As a result, both clients and providers retreat to the safety of FTEs and SLAs, even while acknowledging that these structures do little to drive real value.

GenAI changes the equation.

To unlock its benefits, contract structures must evolve. Value creation can no longer be treated as a vague future promise while cost certainty is locked in today.

The Hard Question No One Wants to Answer

If GenAI creates upside — but also uncertainty — who takes the risk?

Clients have grown accustomed to pushing risk to providers. Providers have grown accustomed to protecting margins through volume, change control, and contractual insulation.

That dynamic will not survive an AI-driven delivery model.

What Must Change

To make GenAI work at scale, three things have to shift:

1. Radical TransparencyProviders will need to become comfortable sharing real productivity data, delivery approaches, and innovation roadmaps. Today, very few providers have a true structural advantage—but with AI, some will. For others, lack of transparency will become a competitive liability.

2. Shared Risk AcceptanceClients must become more comfortable carrying some delivery risk. The days of simply transferring all uncertainty to the provider are ending. AI benefits cannot be guaranteed upfront—and pretending otherwise will stall adoption.

3. Relationship and Trust as Core AssetsAs uncertainty increases, trust becomes more valuable—not less. Some outcomes cannot be fully priced, measured, or governed upfront. They will need to be resolved after results are known, not before contracts are signed.

Practical Guidance: How to Start

This is not a call for wholesale reinvention overnight. Smart organizations will evolve deliberately.

• Start small and build Pilot AI-enabled delivery in less critical areas. Use measurable milestones. Learn before scaling.

  
  

• Keep it simple

  If a gain/risk-sharing construct cannot be explained in two or three pages, it will never be used effectively. Complexity kills accountability.
  

• Build trust before you need it

  Clients want to lock in AI benefits now—even if results come later. The only sustainable way to do this is through transparent sharing of both upside and downside.
  

Key Takeaways

GenAI will not magically fix software delivery.

But it will expose which client–provider relationships are built for value creation—and which are built for risk avoidance.

The future of managed services will not be defined by who jumps first on the AI bandwagon, but by who restructures relationships to make AI worth adopting at all.