Generative AI Crossed the Chasm. Agents Haven’t.

Blogs » Generative AI Crossed the Chasm. Agents Haven’t.

The 2026 Stanford AI Index Report is out and it frames two stories that matter for enterprise leaders. One is largely settled. The other is just beginning.

Generative AI has crossed the chasm
88% of organizations now use AI in at least one business function, up from 78% a year earlier. 79% regularly use generative AI in at least one function, compared with 71% in 2024. China and Europe posted the largest year-over-year jumps in organizational AI use.

The value evidence is keeping pace. Peer-reviewed studies compiled by Stanford show 14–15% more issues resolved per hour in customer support, 26% more pull requests completed by software developers using AI coding assistants, and a 50% increase in output per worker for marketing teams using multimodal AI for ad creation.

The debate about whether generative AI delivers measurable enterprise value is, by the weight of evidence, over.

Agentic AI is the other story
Agentic AI now dominates the tech press, the analyst calendars, and the keynote slides. The Stanford data shows a different reality.

When organizations were asked how far AI agents had moved into production, the answers was in single digits. In manufacturing, 4%. In risk and compliance, 3%. In supply chain, 3%. Even in IT and knowledge management, the leading functions, about two-thirds of respondents report no agent use at all. Only the technology sector breaks the pattern, with scaled agent use of 24% in software engineering and 22% in IT, which makes sense given that the technology sector is the one building the tools.

Why agentic adoption is lagging: our read
The Stanford report documents the gap but doesn’t speculate much on the causes. From our experience, six factors explain most of it:

  • Integration depth: Unlike Generative AI, Agents must plug into systems of record: EHRs, CRMs, ERPs, billing platforms. That is enterprise software engineering, not prompt engineering.
  • Data readiness: A chatbot can converse over messy data and still sound helpful. An agent that acts on that data will compound the mess.
  • Accountability: When an agent submits a claim, books an appointment, or denies a request, who owns that decision for a regulator, an auditor, or a patient?
  • Governance: Beyond accountability for individual actions, what are the standing controls: observability, audit trails, model-change management, human-in-the-loop boundaries, rollback paths?
  • Human comfort with ceding control: The shift from “AI suggests, human acts” to “AI acts” is a cultural change, not just a technical one. Trust must be earned, in increments, over time.
  • The org chart: As work transitions to a predominantly agentic operating model, what should the current org chart become? Which functions consolidate, which expand, which disappear? Where do agents sit, who hires for them, who is accountable for their outcomes, and how do human roles reshape around them?

These are solvable problems. They are not solvable through tooling alone.

The promising signs are in HLS
Healthcare and Life Sciences is one of the most regulated, most data-rich, and most capital-intensive sectors in our economy, and one where AI can impact by: lowering the cost of care, raising the quality of care, and scaling its availability and reach. The impact ceiling is unusually high, which is why the early production evidence matters.

The report documents the scale of AI in HLS today with examples. The framing is direct: “Clinical AI moved from pilot-stage initiatives to enterprise-scale deployments in 2025.” Abridge’s ambient documentation reached 63% of hospitals using Epic and produced a 112% ROI at Northwestern Medicine. Bayesian Health’s TREWS sepsis prediction at Cleveland Clinic delivered an 18.7% relative reduction in sepsis mortality with an 89% clinician adoption rate. OpenEvidence is now used by 40% of U.S. physicians.

These deployments are largely generative and predictive (ambient scribes, alerting, evidence retrieval), with a clinician still in the action loop. They are not yet, in the strict sense, agentic. But they are the proof that HLS has solved the harder problems for AI at scale: deep EHR integration, governance frameworks like Stanford Health Care’s FURM, clinician adoption, and measurable clinical and financial outcomes. That foundation is exactly what the agentic chapter will be built on.

Where Lirik is investing
Lirik’s investment is concentrated in the HLS workflows where agentic AI’s ROI is easiest to achieve while being compliant. Our agents include:

  • Prior Authorization Agent. Automates payer submission, status tracking, and documentation eliminating billions in administrative cost.
  • Denials Management Agent. Categorizes, prioritizes, and drafts appeals, with measurable impact on recovered revenue and AR days.
  • Patient Scheduling Agent. Handles intake, slot optimization, reminders, and no-show recovery to improve utilization and availability in real-time.
  • Value-Based Care Coordination Agent. Surfaces care gaps and align outreach to quality measures and shared-savings contracts.

Alongside the agents themselves, we are investing in the data layer that makes both agentic and generative AI work in HLS. Our FHIR-to-OMOP mapping accelerators bridge the operational world of clinical events with the analytical world. This is foundational in addressing the data readiness barrier by breaking down silos and giving agents and generative AI a clean, governed intelligence layer to feed from. If you’d like to learn more about data mapping, or watch an agentic demo, or ask about any other use cases, contact us.

Source: All data and figures cited in this article are drawn from The AI Index 2026 Annual Report, Stanford Institute for Human-Centered Artificial Intelligence (HAI), April 2026.

Lirik empowers businesses to seize global opportunities with top-tier CRM, ERP, and data solutions. We combine startup agility with enterprise maturity, delivering personalized experiences, operational excellence and transformative growth.

Talk to one of our experts.

If you are applying or looking for a job, please email hiring@lirik.io.

    Global Delivery Centers

    Gurgaon

    Fortune Towers II, Floor #5 406 Udyog Vihar, Phase III Gurgaon, India, 122016

    Noida

    Vertex Tower, Plot no-
    C-33, 4th Floor, Phase 2, Industrial Area, Sector 62, Noida, Uttar Pradesh, 201309

    Pune

    Suma Center, 6th Floor Near Deenanath Mangeshkar Hospital Pune, India, 411004

    Jaipur

    IndiQube Fort, 3rd Floor Malviya Nagar, Jaipur India, 302017

    Nagpur

    4th Floor, JK Heights
    Ajni Square, Deo Nagar Nagpur, India, 440015

    Copyright 2026 Lirik All Rights Reserved.

    Copyright 2026 Lirik All Rights Reserved.