The need for matched normal in variant-calling: gene-panel vs. whole exome

Hi 🙂

1) Somatic SNV calling of whole-exome data by tools such as Mutect require a matched-normal.
Why is this not a requirement for variant-calling in products like this one? (Avenio, targeted capture kit of 77 genes, hybrid-capture, 92kbases). How is the magic done by such proprietary variant-callers?

2) I have the BAM file from the above kit and I would like to do the variant-calling myself.
Can I use Mutect for such a small target capture (92kbase)? What can I use as panel of normals?
Or would you recommend another tool for the task?

Any clue would be appreciated!