With dozens of telescopes in both hemispheres, the Cherenkov Telescope Array Observatory (CTAO) will be the largest ground-based gamma-ray observatory and will offer extensive energy coverage from 20 GeV to 300 TeV. Its large effective area, wide field-of-view, rapid slewing capability, and exceptional sensitivity will make CTAO an essential instrument for the future of ground-based gamma-ray astronomy. Furthermore, its two arrays will send alerts on transient and variable phenomena (e.g., gamma-ray bursts, active galactic nuclei, gamma-ray binaries, and serendipitous sources) to maximise the scientific return. Effective and rapid communication with the community requires a reliable automated system to detect and issue candidate science alerts. This automation will be achieved by the Science Alert Generation (SAG) pipeline, a core system of the CTA Observatory. The SAG is part of the Array Control and Data Acquisition (ACADA) system. The SAG working group develops pipelines for data reconstruction, data quality monitoring, science monitoring, and real-time alert issuance to the Transients Handler system of ACADA. The SAG performs the first real-time scientific analysis during data acquisition. The system analyzes data on multiple time scales (from seconds to hours) and must issue candidate science alerts with 20 seconds of latency and at least half the CTAO nominal sensitivity. Dedicated, highly optimized software and hardware architectures must be designed and tested to satisfy these stringent requirements and manage trigger rates of tens of kHz from both arrays. In this work, we present the general architecture and current development status of the ACADA/SAG system.