AI prompts for DCFs
Practitioners are standardizing AI prompts to build unlevered DCFs with revenue growth curves, margin progression, FCF schedules, WACC, terminal value and sensitivity tables—then producing bull/base/bear cases from consensus inputs. The playbook emphasizes auto‑pulling consensus data and creating scenario matrices for quick valuation comparisons .
An open-source valuation toolkit named dcfvaluation on GitHub implements a full DCF framework with structured data gathering, parallel WACC and projection calculations plus scenario modeling. (github.com) Prompt marketplaces such as PromptBase list thousands of finance templates and report a catalog of about 260,000 prompts across formats and use-cases. (promptbase.com) Standalone GPTs and specialized agents have been built to run DCF workflows: ValueBot is offered as a GPT focused on automating DCF analysis and sourcing inputs from the web. (theresanaiforthat.com) Consensus inputs are routinely pulled from vendor APIs in production pipelines: FactSet’s [Estimates API] exposes rolling- and fixed-consensus endpoints and enforces a 10 requests/second rate limit. (developer.factset.com) Refinitiv/LSEG’s I/B/E/S Estimates API provides analyst forecast coverage back decades and aggregates contributor-level estimates used for model calibration. (developers.lseg.com) Spreadsheet AI features are being wired into these prompts: Microsoft’s Copilot in Excel added Agent Mode and a =COPILOT() function to generate formulas, scenario analyses and multi-step workflows inside workbooks. (techcommunity.microsoft.com) Guides testing Copilot report it can cut repetitive formula and chart-building time by roughly 60–80% in financial workflows. (aiacopilot.com) OpenAI’s Custom GPTs let teams bake DCF instructions, file attachments and tool calls into reusable assistants for repeatable valuation work. (academy.openai.com) Concrete prompt-driven implementations already use formal scenario matrices: Dave Wang’s Circle DCF example encodes projection_years: 3 and explicit bull/base/bear discount_rate_pct values (base 15.0%, bull 13.0%, bear 17.0%) along with exit EV/Revenue multiples of 5/6/4 in YAML-style scenario blocks. (davewang.ai) Regulators are watching the shift: the SEC flagged increased examinations of AI tools used by brokers and investment advisers in October 2024, and the UK FCA published an AI approach update on April 22, 2024 that embeds AI oversight into existing conduct and model-risk frameworks. (cpapracticeadvisor.com)
