Q: What types of AI tools do venture capitalists actually use day to day? A: Investors mainly use transcription and note-taking tools, large language models and research/data platforms, and early agent workflows to automate repetitive tasks. Email automation still lags and is used sparingly because it often creates extra work. Q: Which transcription and note-taking tools are most common among investors? A: Investors cited Granola for its ability to capture conversational nuance and Otter for its mature, widely adopted transcripts. They often follow up with a quick human pass to tag action items and share summaries into the CRM. Q: How do investors use large language models like ChatGPT and Claude? A: Investors use Claude and ChatGPT for general analysis, brainstorming, drafting memos, and as “thought partners” on new sectors, while reserving more complex reasoning for models like Strawberry. They deliberately choose models by task and validate outputs against data platforms before adding conclusions to pipelines. Q: What platforms help with sourcing and tracking early-stage companies? A: Specter and Harmonic are used for sourcing and tracking early-stage companies because they pull structured company signals, and Strawberry is used when deeper research and citations are needed. Teams typically export clean fields from these platforms into their CRM for pipeline hygiene. Q: Are AI agents ready for widespread use in venture firms? A: Some investors are experimenting cautiously with agent platforms such as Lua AI, Cognosys, and Adept to automate tasks like web scraping, populating spreadsheets, and running quick scenario checks. Expectations remain measured: Gartner predicts more than 40% of agentic AI projects will be canceled by the end of 2027, and IT surveys flag security issues such as agents touching unintended systems or using inappropriate data. Q: Why do many investors avoid using AI to write or send emails? A: Email copilots often produce “workslop”—machine-generated content that colleagues must fix—so investors say rewrites can take longer than writing from scratch. A Stanford study found 40% of workers had encountered such workslop, costing nearly two hours to resolve and about $186 per worker per month, which explains reluctance to fully automate emails. Q: What guardrails should firms put in place when piloting agentic AI? A: Guardrails include human-in-the-loop checks for any external message or purchase, least-privilege access with sandboxes for tests, full logging, cost ceilings and kill switches, and data redaction for PII and sensitive deal notes. Firms should also use enterprise plans that specify data handling, retention limits, and prompt redaction. Q: How should a firm pick and trial the best AI tools for venture capitalists? A: To pick the best AI tools for venture capitalists, start small with three use cases—meeting notes, sector research, and sourcing—then run 30-day trials with a small pod of users. Keep tools that save at least 20% time with equal or better quality, test models by task, and enforce enterprise data controls such as no-training clauses and retention limits.