To solve the given optimization problem, we first need to translate the natural language description into a symbolic representation. This involves defining variables for the quantities of each supplement the man should take and then formulating the objective function and constraints based on these variables.

Let's define:
- $x_1$ as the number of servings of supplement A
- $x_2$ as the number of servings of supplement B

The natural language objects and their corresponding symbolic variables are:
```json
{
    'sym_variables': [('x1', 'servings of supplement A'), ('x2', 'servings of supplement B')],
    'objective_function': 'Minimize 2*x1 + 4*x2',
    'constraints': ['x1 + 4*x2 >= 12', '5*x1 + x2 >= 20', 'x1 >= 0', 'x2 >= 0']
}
```

The objective function `2*x1 + 4*x2` represents the total cost, which we want to minimize. The constraints are:
- `x1 + 4*x2 >= 12`: Ensures that the man gets at least 12 units of keratin.
- `5*x1 + x2 >= 20`: Ensures that he gets at least 20 units of calcium.
- `x1 >= 0` and `x2 >= 0`: Non-negativity constraints, meaning he cannot take a negative number of servings.

Now, we can write the Gurobi code to solve this linear programming problem:

```python
from gurobipy import *

# Create a new model
m = Model("Supplement_Optimization")

# Define variables
x1 = m.addVar(name='x1', lb=0)  # Servings of supplement A
x2 = m.addVar(name='x2', lb=0)  # Servings of supplement B

# Set the objective function
m.setObjective(2*x1 + 4*x2, GRB.MINIMIZE)

# Add constraints
m.addConstr(x1 + 4*x2 >= 12, name='keratin_requirement')
m.addConstr(5*x1 + x2 >= 20, name='calcium_requirement')

# Optimize the model
m.optimize()

# Print the solution
if m.status == GRB.OPTIMAL:
    print(f"Servings of supplement A: {x1.x}")
    print(f"Servings of supplement B: {x2.x}")
    print(f"Total cost: {m.objVal}")
else:
    print("No optimal solution found")
```