TAM

$298.2 billion

TAM defined as the global annual software spend where a continuous agent-improvement platform would be purchased (enterprise AI application software + AI infrastructure software). Gartner’s 2025 AI spending breakdown lists AI Application Software at $172.029 billion and AI Infrastructure Software at $126.177 billion for 2025; summing those line items produces a 2025 software-focused addressable TAM of ~$298.206 billion. This definition intentionally focuses on the software layers (platforms, application software, infrastructure software) where procurement decisions for automated agent-improvement, prompt-generation, model-retraining automation, and deployment/observability tooling happen; it excludes large hardware and consumer-device line items to avoid double-counting. Adjacent specialist markets (MLOps/model operationalization and generative-AI application growth) show rapid expansion that reinforces this software TAM perspective.

• gartner.com
• grandviewresearch.com
• mordorintelligence.com

SAM

$40.8 billion

SAM defined as the subset of the TAM that is directly addressable by a platform that continuously improves production AI agents: (1) AI/ML observability & full‑stack observability (model and telemetry monitoring for production agents), (2) MLOps / model-operationalization tooling, and (3) production agent / chatbot application software (customer-facing and internal agents). Using recent market estimates for 2025 yields: observability tools & platforms ≈ $28.18B (2025 projection), MLOps ≈ $3.03B (2025 estimate), and chatbot/agent application software ≈ $9.56B (2025 estimate); these three categories sum to ≈ $40.77B, rounded to $40.8B. Rationale: these segments are where teams buy monitoring, root-cause analysis, retraining workflows, and prompt/agent tuning — i.e., the direct procurement use-cases for an automated agent self-improvement product. The SAM intentionally excludes broader AI application & infrastructure line items in the TAM to reduce overlap and to focus on the near-term commercial opportunity for specialized agent-improvement tooling.

• marketresearchfuture.com
• grandviewresearch.com
• grandviewresearch.com

SOM

$204 million

SOM estimated as a realistic early‑go‑to‑market capture over a multi‑year rollout (3–5 years) focused on enterprise and mid‑market customers within the SAM. Method: apply a conservative attainable penetration of 0.5% of the $40.77B SAM (0.005 × $40.77B = $203.85M), rounded to $204M. This assumes a targeted enterprise GTM (direct sales + land-and-expand), initial vertical focus (e.g., finance, e‑commerce, SaaS platforms), and multi‑year adoption to reach scale. Unit-economics sanity check: using a representative enterprise ACV assumption of $150k (enterprise AI tooling ACVs vary widely; private-SaaS median ACV is ~$26k overall while enterprise-focused deals commonly run materially higher, consistent with industry benchmarks), hitting $204M requires roughly 1,360 paying customers at that ACV (203,850,000 / 150,000 ≈ 1,359). The $150k ACV is an explicit assumption for an enterprise GTM; median-private SaaS ACV benchmarks and industry surveys provide context for this assumption. Sensitivity: an instrumented scenario range is provided by simple scaling — 0.25% SAM → ~$102M (lower-case), 1.0% SAM → ~$408M (upside). The chosen 0.5% SOM is a conservative, investor-style early‑market estimate consistent with a focused enterprise GTM and expects expansion revenue and upsells to drive growth after initial deployments.

• grandviewresearch.com
• marketresearchfuture.com
• grandviewresearch.com
• saas-capital.com