Can Legacy Automation Vendors Adapt to Agentic AI? PEGA Adopts the 'Harness' Framework.
One of the analytical dilemmas in software right now is whether legacy automation vendors (RPA) can adapt to the agentic AI era or whether agentic AI simply hollows them out.
PEGA is a useful case study. It is not a pure-play RPA vendor in the narrow sense, and PATH will likely be the cleaner test. But PEGA sits in the same broad problem space: enterprise workflow automation, decisioning, orchestration, and process execution. Management is now explicitly trying to redefine itself as the governed execution layer for enterprise AI.
The repositioning came through clearly in the latest earnings call. The company's new language is built around the "harness" concept. This is a concept gaining pupularity among AI professionals lately. PEGA's Blueprint would be the design-time harness that uses AI up front to map, redesign, and structure work. Pega Platform / Infinity is the run-time harness that executes those workflows with governance, auditability, orchestration, and selective model use. In management's framing, ungoverned runtime AI is problematic: expensive, inconsistent, hard to test, and dangerous in regulated environments. That is a real strategic response to the agentic AI threat. This is not really new defence for RPA vendors but now the AI community itself provides the arguments by stating that "agents performance incresingly depends on harness engineering". So PEGA did not invent the term either but it immediately adopted it. Honestly, based on my own experience in using AI, the harness really matters.
This is why the latest earnings analysis treated the "harness" language as more than branding. CEO Alan Trefler spent time defending the relevance of enterprise software itself against LLMs, vibe-coding, and the idea that AI-generated code can replace architecture. The problem is that the numbers are not yet proving the strategic argument cleanly enough.
This is where the PEGA case gets interesting. If you go back through the previous earnings results, the first part of the story looked constructive. The Q1-Q3 2025 analysis saw genuine acceleration in the cloud engine. Pega Cloud ACV was improving, Pega Cloud revenue was accelerating, and Blueprint seemed to be changing the sales motion. The Q4 2025 analysis kept that generally positive posture, but it also flagged the key tension: total ACV growth in constant currency was not really accelerating, even as management was heavily promoting Blueprint and AI momentum. In other words, the cloud story was getting better, but the overall business was not clearly following it.
The Q1 2026 earnings pushed that tension much further. It moved from a constructive reading to a deceleration. The reason was straightforward. Constant-currency total ACV growth, which had already been flat at 14% for three quarters, dropped to 11% in Q1 2026. Net new ACV slowed sharply. The sequential ACV add fell to just $14 million, down from $43 million in Q3 2025 and $51 million in Q4 2025. Backlog growth also softened, posting the weakest print in the sequence. Those are not cosmetic issues. Those are the leading indicators investors should care about.
And yet the cloud business still looks healthy. Pega Cloud ACV kept growing, reaching roughly $907 million in Q1 2026. Pega Cloud revenue rose to about $205 million and was still the one clearly accelerating line item, growing 36% year over year in Q1. That is what makes the PEGA debate so analytically useful. The company appears to be doing something right in cloud and in AI-enabled go-to-market, but cloud growth still does not seem strong enough to offset the deceleration in the broader business. Maintenance erosion, term-license runoff, and weaker total ACV momentum are absorbing too much of the benefit.
That is the crucial lesson from this case study. A legacy automation vendor may be able to adapt strategically to the agentic AI era before it adapts financially. PEGA may have identified the right narrative and even the right product architecture, but that does not automatically translate into a reaccelerating total-company growth profile. The market still needs evidence that the new AI-control-layer positioning will produce enough incremental demand to outweigh the drag from the legacy parts of the model.
Management's own guidance makes that tension hard to ignore. FY2026 ACV growth guidance was set at 15%, but Q1 came in at 11% in constant currency. Pega Cloud ACV growth had been guided to 30%+, but Q1 printed 27% after reaching 28% in Q4. Management says the year is back-end loaded, the renewal cycle is skewed toward the second half, and Blueprint-driven pipeline should begin converting later in 2026. That may all be true. But it means investors are no longer looking at a story that is obviously accelerating. They are looking at one that requires trust in future conversion.
PEGA shows that incumbents are not trapped with only one option, which is to deny the AI threat. There is a viable adaptation path: become the design-time and run-time control layer around AI, lean into governance and orchestration, and make your value proposition about reliable enterprise execution rather than just software development. PEGA is clearly attempting that move, and the logic behind it is sound. But PEGA also shows the harder truth. Even if the strategic adaptation is correct, the financial proof can lag for a long time. Cloud can accelerate while the total business still decelerates. A better AI narrative can coexist with weaker leading indicators. Legacy runoff can neutralize next-generation momentum longer than bulls expect.
PATH may give us the cleaner RPA version later. But PEGA already shows what the debate looks like when the strategy is plausible and the numbers are still catching up.