With the development of financial technology, how to integrate modern credit technology to effectively alleviate the financing difficulties of small and micro enterprises is a key problem to be solved in the financial market. From the perspective of “interaction of banks and taxes” policy, this paper excavates the credit characteristics of tax information of small and micro enterprises, and evaluates the credit default probability and credit risk level of small and micro enterprises in combination with random forest method and deep neural network method. Then, the profit and cost of the bank are comprehensively considered to optimize the credit interest rate and credit limits of different small and micro enterprises, so as to build an intelligent decision-making system for credit supply of small and micro enterprises. The results show that tax information can accurately identify the credit risk of small and micro enterprises, thus providing decision-making basis for bank credit provision. Furthermore, small and micro enterprises with high quality of tax information usually obtain more favorable credit support in whether credit limits or credit interest rate.