This study examines whether large language models (LLMs) can effectively assist investors in incorporating venture capital (VC) investments into their portfolios. Using 48 hypothetical investor profiles that vary in VC focus, investor status, investment horizon, risk tolerance, and home country, we elicit portfolio recommendations from four reasoning LLMs. We find that LLMs incorporate VC preferences by increasing allocations to VC-like investment products and decreasing allocations to public equity and fixed income. LLMs recommend larger VC allocations to accredited investors than to retail investors. VC-focused prompts generate portfolios that mirror the more aggressive risk–return characteristics of VC investments. These portfolios exhibit higher exposure to the Fama–French size factor, lower exposure to the investment factor, higher historical excess returns and alphas, and additional risk that is primarily systematic. Overall, our results suggest that LLMs can incorporate nuanced investment objectives, potentially assisting investors with VC portfolio construction and broadening retail investors’ access to VC-style investments.