The fastest athletes in the world can botch a baton pass if trust isn’t there—and the same is true of AI, Blackbaud exec says

Speakers at Fortune’s Brainstorm AI argued that the primary barrier to commercial AI adoption is not raw model speed but trust: Blackbaud CPO Sudip Datta likened the gap to relay teams that fumble batons, asserting that trust is a revenue driver rather than a compliance drag. LiveRamp CEO Scott Howe outlined five conditions for trust—transparency, control, fair exchange of value, portability and interoperability—saying regulation has helped the first two but big tech resistance to portability and interoperability has created data silos and leaves consumers shortchanged. Enterprise deployments also face practical trust challenges—what AI can operate autonomously, hallucinations, cybersecurity and critical-infrastructure risk—and vendors such as ServiceNow are using orchestration layers and specialized small models to limit exposure; firms that can credibly solve these trust and governance issues are positioned to accelerate adoption and capture value, while regulatory and data-portability dynamics remain key risks to monitor.

Analysis

Speakers at Fortune’s Brainstorm AI framed a central commercial constraint on AI adoption as trust rather than raw model speed, with Blackbaud CPO Sudip Datta using a 4×100-relay analogy to argue that “trust is actually a revenue driver” because lack of faith at key handoffs causes innovation to fumble. LiveRamp CEO Scott Howe outlined five conditions for trust—transparency, control, exchange of value, portability and interoperability—and said regulation such as the GDPR has advanced the first two but the industry is “nowhere on” portability and interoperability, creating data silos that entomb customers in incumbent ecosystems.

The article cites major tech platforms (Amazon, Google, Walmart) as resisting portability and interoperability, a stance that increases antitrust and regulatory risk and leaves consumers and downstream enterprise customers with limited bargaining power. ServiceNow’s Spencer Beemiller described practical enterprise risk-management responses: orchestration layers that route queries to specialized small models for precision and larger models for conversational tasks, and monitoring of agent lifecycles to decide what can run autonomously.

Key unresolved risks highlighted include hallucinations, cybersecurity and critical-infrastructure exposures, and the governance question of which AI agents can operate without human oversight; vendors that can demonstrably solve these trust, portability and governance problems are positioned to accelerate adoption and capture value, while regulatory shifts on portability/interoperability will materially re-rate incumbents and specialists differently.

Key Decisions for Investors

• Consider increasing exposure to enterprise software vendors that explicitly monetize trust and governance features and offer orchestration or data-control capabilities (examples cited include BLKB, RAMP, NOW), as these firms are positioned to capture adoption if their solutions prove effective
• Monitor regulatory developments on data portability and interoperability closely and consider hedging or reducing exposure to large platform incumbents that resist portability (AMZN, GOOGL, WMT) because adverse rulings could erode their ecosystem advantages
• Require demonstrable, revenue-linked KPIs before adding positions in AI plays: look for customer contract wins, enterprise deployment metrics, SLAs around hallucination mitigation and agent oversight, and be prepared to reallocate capital if those trust/governance proofs do not materialize