[guided]Fix a point $x \in M$. The definition of $b_\rho(x)$ requires a limit as the parameter point $\rho(t)$ moves out along the ray, so we first prove that the real-valued function being limited has a finite limit. Define
\begin{align*}
F_x: [0,\infty) &\to \mathbb{R} \\
t &\mapsto t - d_g(x,\rho(t)).
\end{align*}
The key point is that increasing $t$ moves farther along a minimizing ray. Let $0 \le t_1 \le t_2$. Since $\rho$ is a ray, the segment from $\rho(t_1)$ to $\rho(t_2)$ has distance exactly equal to its parameter length:
\begin{align*}
d_g(\rho(t_1),\rho(t_2)) = t_2 - t_1.
\end{align*}
The triangle inequality gives
\begin{align*}
d_g(x,\rho(t_2))
&\le d_g(x,\rho(t_1)) + d_g(\rho(t_1),\rho(t_2)) \\
&= d_g(x,\rho(t_1)) + t_2 - t_1.
\end{align*}
After moving terms to the other side, this becomes
\begin{align*}
t_1 - d_g(x,\rho(t_1)) \le t_2 - d_g(x,\rho(t_2)).
\end{align*}
Thus $F_x(t_1) \le F_x(t_2)$ whenever $t_1 \le t_2$, so $F_x$ is monotone increasing.
A monotone increasing function has a finite limit if it is bounded above. The upper bound comes from the reverse triangle inequality. Applied to $x$, $\rho(0)$, and $\rho(t)$, it gives
\begin{align*}
d_g(x,\rho(t)) \ge d_g(\rho(0),\rho(t)) - d_g(x,\rho(0)).
\end{align*}
Because $\rho$ is a ray starting at $\rho(0)$,
\begin{align*}
d_g(\rho(0),\rho(t)) = t.
\end{align*}
Therefore
\begin{align*}
F_x(t)
= t - d_g(x,\rho(t))
\le t - \bigl(t - d_g(x,\rho(0))\bigr)
= d_g(x,\rho(0)).
\end{align*}
So $F_x$ is monotone increasing and bounded above by the finite number $d_g(x,\rho(0))$. Hence
\begin{align*}
\lim_{t \to \infty} F_x(t)
\end{align*}
exists in $\mathbb{R}$. This proves that $b_\rho(x)$ is well-defined for the chosen point $x$, and since $x \in M$ was arbitrary, $b_\rho$ is defined on all of $M$.[/guided]