The centrality of pyruvate oxidative decarboxylation into acetyl-CoA in current biochemistry is a strong argument for proposing that a similar reaction have been necessary for the development of an effective protometabolism on the primitive Earth. However, such a decarboxylation requires the use of an oxidant and a catalyst, today enzymatic. Based on the mechanisms of the pyruvate dehydrogenase complex and pyruvate-ferredoxin oxidoreductase, we propose that the initial mechanism involved disulfides and occurred via radicals. A first disulfide is obtained by reacting glyoxylate with hydrogen sulfide. It is then possible to produce a wide variety of other disulfides by exchange reactions. When reacted with pyruvate under UV light they give thioesters. This process requires no oxidant and is therefore compatible with what is known of the redox conditions of the early Earth. Neither does it require any catalyst. It could be the first way to acetyl thioesters, a way that was later improved by the introduction of catalysts, first minerals, then enzymes.