Conventional handwriting recognition systems often fail to capture complex hand-device interactions, such as grip force distribution and subtle posture changes, but remain susceptible to hand-tremor noise. To overcome such inherent limitations of traditional single-point pressure-sensing systems, we developed a pressure-array-based manuscript optimization (PAMO) system with an intelligent stylus integrated with a 128-unit flexible piezoresistive sensor array (16 × 8 units) that captures a high-dimensional manifold of grip force data at a 120 Hz sampling frequency. The PAMO system conducts pressure-adaptive handwriting optimization using a dual-layer optimization framework and includes a second-order Kalman filter for real-time denoising and position prediction and Cubic B-spline interpolation to ensure G2 and C2 continuity. The PAMO system significantly improves trajectory smoothness, achieving a 42.7% reduction in the standard deviation of curvature for straight lines (0.201 vs 0.351 for the baseline system) and a 42.0% reduction for circles (0.283 vs 0.488 for the baseline system). The one-step prediction function of the PAMO system reduced end-to-end latency to 16.9 ms, showing an 8.5 ms improvement over the baseline system. By mapping dynamic grip features, such as total grip force, center of pressure, and pressure gradient, to morphological stroke parameters, the PAMO system represents a significant advancement toward approximating the nuanced physical reality of traditional artistic expression.