Cells are essentially living computers, which they receive, integrate and respond to various signals. Although sharing an origin from a single zygote and a genome encoding the same genes, they differentiate into a myriad of cell types/states, with different functions. These differences may derive from different signals or heterogeneous responses to the same signal. Capturing the history of cells can help us understand their current state and predict or even shape their future behavior. Current methods of profiling cell states usually require destruction and only provide a snapshot of cell states, losing the information about their history. Can we have a device in the cells that runs autonomously to record cellular signals and events? Recently engineered genome engineering tools-CRISPR/Cas9 and its derivatives- provide a unique platform for such a purpose, allowing researchers to effectively alter/rewrite DNA sequence in a highly specific manner. In my thesis, I focused on developing a DNA-based memory system where molecular signals and events could be recorded by alternating a pre-programmed stretch of DNA sequence, which we termed DNA TAPE. I describe experiments and methods to mainly address two questions: 1. How information can be encoded with the CRISPR system. 2. What information we can record. We investigated the possibility of encoding information with the CRISPR system with three different approaches. We first used CRISPR/Cas9 (cut) to introduce pseudo-random mutations and built a machine learning model (Lindel) to accurately predict the mutations and their frequencies. For each target in a DNA TAPE, this approach allows us to encode ∼3 bits of information (8 different states). We next used prime editing to introduce large deletions (Prime-del) or short insertions (ENGRAM), which allows us to encode 1 bit and up to 20 bits of information per target in DNA TAPE. We next sought to record various signals in cells. As most of the cell functions are executed by transcription, we recorded transcription activities of hundreds of enhancers and 3 different signals TetOn, NFκB and Wnt. We show that many signals can be recorded simultaneously with reasonable efficiency in a digital manner. We also discussed future applications of recorders in understanding different biological functions.