diff --git a/2023/Cargo.toml b/2023/Cargo.toml
index 7000a92..877cfdf 100644
--- a/2023/Cargo.toml
+++ b/2023/Cargo.toml
@@ -10,6 +10,7 @@ aho-corasick = "1.1.2"
 anyhow = "1.0.75"
 clap = { version = "4.4.8", features = ["derive"] }
 nom = "7.1.3"
+num-integer = "0.1.45"
 
 [dev-dependencies]
 criterion = "0.5.1"
diff --git a/2023/benches/days.rs b/2023/benches/days.rs
index c85789a..622a818 100644
--- a/2023/benches/days.rs
+++ b/2023/benches/days.rs
@@ -9,7 +9,7 @@ use criterion::Criterion;
 use aoc_2023::get_implementation;
 
 /// Number of days we have an implementation to benchmark
-const DAYS_IMPLEMENTED: u8 = 7;
+const DAYS_IMPLEMENTED: u8 = 25;
 
 fn read_input(day: u8) -> std::io::Result<Vec<u8>> {
     let input_path = format!("inputs/{day:02}.txt");
diff --git a/2023/src/day08.rs b/2023/src/day08.rs
index 1c60e39..c93e44e 100644
--- a/2023/src/day08.rs
+++ b/2023/src/day08.rs
@@ -1,12 +1,15 @@
+use anyhow::Context;
 use nom::bytes::complete::tag;
 use nom::bytes::complete::take;
 use nom::bytes::complete::take_until;
 use nom::combinator::map;
 use nom::multi::fold_many1;
+use nom::sequence::preceded;
 use nom::sequence::separated_pair;
 use nom::sequence::terminated;
 use nom::sequence::tuple;
 use nom::IResult;
+use num_integer::Integer;
 
 use crate::common::parse_input;
 
@@ -62,6 +65,25 @@ fn parse_map(i: &[u8]) -> IResult<&[u8], Map<'_>> {
     )(i)
 }
 
+fn parse_starts(i: &[u8]) -> IResult<&[u8], Vec<u16>> {
+    preceded(
+        tuple((take_until("\n"), tag("\n\n"))),
+        fold_many1(
+            terminated(
+                map(take(3usize), place_to_index),
+                tuple((take_until("\n"), tag("\n"))),
+            ),
+            Vec::new,
+            |mut starts, place| {
+                if place % 26 == 0 {
+                    starts.push(place)
+                }
+                starts
+            },
+        ),
+    )(i)
+}
+
 pub fn part1(input: &[u8]) -> anyhow::Result<String> {
     let map = parse_input(input, parse_map)?;
     let end = place_to_index(b"ZZZ");
@@ -78,8 +100,32 @@ pub fn part1(input: &[u8]) -> anyhow::Result<String> {
     anyhow::bail!("Unreachable, loop is infinite");
 }
 
-pub fn part2(_input: &[u8]) -> anyhow::Result<String> {
-    anyhow::bail!("Not implemented")
+// This code is wrong. There is no reason that the start of the cycle is indeed the equal to the
+// length of the cycle. But it happens to be the case, so we roll with it. Otherwise you could go
+// with the full Chinese remainder theorem and knock yourself out that way.
+//
+// I didn't wanna.
+fn find_cycle(map: &Map<'_>, start: u16) -> usize {
+    let mut pos = start;
+    for (count, &step) in map.instructions.iter().cycle().enumerate() {
+        if pos % 26 == 25 {
+            return count;
+        }
+        pos = map.transition(pos, step);
+    }
+
+    unreachable!("Loop is actually infinite")
+}
+
+pub fn part2(input: &[u8]) -> anyhow::Result<String> {
+    let map = parse_input(input, parse_map)?;
+    let pos = parse_input(input, parse_starts)?;
+
+    pos.iter()
+        .map(|&p| find_cycle(&map, p))
+        .reduce(|a, b| a.lcm(&b))
+        .map(|s| s.to_string())
+        .context("No starting points somehow")
 }
 
 #[cfg(test)]
@@ -88,10 +134,18 @@ mod tests {
 
     const SAMPLE: &[u8] = include_bytes!("samples/08.1.txt");
     const SAMPLE2: &[u8] = include_bytes!("samples/08.2.txt");
+    // N.B. sample modified because I don't want to change my parser logic to deal with ascii digits
+    // in addition to capitals. 1 has been replaced with D, 2 has been replaced with E.
+    const SAMPLE3: &[u8] = include_bytes!("samples/08.3.txt");
 
     #[test]
     fn sample_part1() {
         assert_eq!("2", part1(SAMPLE).unwrap());
         assert_eq!("6", part1(SAMPLE2).unwrap());
     }
+
+    #[test]
+    fn sample_part2() {
+        assert_eq!("6", part2(SAMPLE3).unwrap());
+    }
 }
diff --git a/2023/src/samples/08.3.txt b/2023/src/samples/08.3.txt
new file mode 100644
index 0000000..a658875
--- /dev/null
+++ b/2023/src/samples/08.3.txt
@@ -0,0 +1,10 @@
+LR
+
+DDA = (DDB, XXX)
+DDB = (XXX, DDZ)
+DDZ = (DDB, XXX)
+EEA = (EEB, XXX)
+EEB = (EEC, EEC)
+EEC = (EEZ, EEZ)
+EEZ = (EEB, EEB)
+XXX = (XXX, XXX)