Panel
Architecting the Agentic Enterprise | Traversal, Kong, Pigment & Twelve Labs | RAISE 2026
- In 2026, the primary focus for enterprises shifts to the reliable utility of agents within enterprise settings rather than general capability, with agents expected to scale safely, securely, and autonomously without functioning as shadow AI systems.
- Digital native companies are predicted to deploy agents into production faster than large enterprises, while the industry transitions from "human in the loop" to "human on the loop" and eventually "human off the loop" for non-core, non-critical use cases.
- System requirements dictate 95% operational reliability with a governance layer managing the remaining 1% to 5% of failures to avoid human distraction and mitigate reputational risks in financial services and regulatory risks associated with unvalidated autonomous actions.
- Token spending is expected to grow significantly as organizations move from incident response to daily production use cases, with costs projected to drop to electricity plus a small margin within the next couple of years due to the emergence of cheaper open-weight models offering a 10x cost decrease compared to frontier models.
- Pricing models are forecasted to split between expensive frontier models and cheaper alternatives, shifting toward fixed-outcome, utility-based, and "intelligence per dollar" metrics resembling large AWS-style commitments to ensure customer predictability.
- Business value is expected to emerge only when agents operate in the core of business value rather than at the edge, with platforms like Pigment aiming to enable teams to become 80% more productive through autonomous planning.
- A high failure rate is anticipated, with 40% of AI initiatives started in the last six months predicted to cease operations in 2027 due to a failure to generate business value, leaving only 5% of companies able to attest to AI contributing to their bottom line or EBIT.
- Responsibility for agents in core business-critical use cases is expected to shift to the business side where value is created, supported by new "transformation roles" that are predicted to disappear as AI integrates into business teams.
- Future workforce capabilities include the emergence of "empiricist" users who research unpredictable use cases and the ability for anyone to become an agent manager as end-to-end workflows become runnable by non-technical staff within three to five years.
- In specific sectors, autonomous driving models are predicted to outperform driver-based models, with Waymo expected to win over Uber based on this fundamental operational difference.