# Sarah Guo’s LAB Benchmark, Reid Hoffman’s Future-Fiction Picks, and Paul Graham on Churchill *By Recommended Reading from Tech Founders • May 7, 2026* Sarah Guo made the clearest practical recommendation of the day with Harvey’s LAB benchmark for legal-agent evaluation and deployment. The rest of the list mixed Reid Hoffman’s future-facing fiction picks, Garry Tan’s endorsement of a Lulu Meservey interview, and Paul Graham’s Churchill productivity cue. ## Why LAB led today’s list The clearest recommendation was also the most practical. Sarah Guo did not just name a resource; she explained what it is for: a long-horizon, open-source legal agent benchmark that can help teams assess what legal agents can do now, plan deployment, and design human-agent cooperation [^1]. ### LAB - **Content type:** Open-source legal agent benchmark / article - **Author/creator:** Harvey [^1] - **Link/URL:** [http://x.com/i/article/2051782974098886656](http://x.com/i/article/2051782974098886656) [^2] - **Who recommended it:** Sarah Guo [^1] - **Key takeaway:** Guo called LAB the first long-horizon, open-source legal agent benchmark and said it helps legal teams answer “what can legal agents do today?”, plan deployment, and design human-agent cooperation [^1] - **Why it matters:** This was the most concrete, deployment-oriented recommendation in the set, with a clear role in evaluating agents inside a hard domain [^1] > "LAB is the first long-horizon, open-source legal agent benchmark... it will help legal teams answer ‘what can legal agents do today?’, plan deployment, and design human-agent cooperation." [^1] ## Worldview builders ### The Culture series - **Content type:** Science fiction books - **Author/creator:** Iain M. Banks [^3] - **Link/URL:** No direct resource URL was provided; source context: [https://www.youtube.com/watch?v=brscUjmA2To](https://www.youtube.com/watch?v=brscUjmA2To) [^3] - **Who recommended it:** Reid Hoffman [^3] - **Key takeaway:** Hoffman pointed to the series as a vision of how AI and human beings can help create the future together [^3] - **Why it matters:** Of Hoffman’s picks, this was the most explicit AI-and-human futures recommendation [^3] [![How Beeple's $69 MILLION sale was just the beginning.](https://img.youtube.com/vi/brscUjmA2To/hqdefault.jpg)](https://youtube.com/watch?v=brscUjmA2To&t=3416) *How Beeple's $69 MILLION sale was just the beginning. (56:56)* ### Murderbot - **Content type:** TV series - **Author/creator:** Not specified in the provided notes - **Link/URL:** No direct resource URL was provided; source context: [https://www.youtube.com/watch?v=brscUjmA2To](https://www.youtube.com/watch?v=brscUjmA2To) [^3] - **Who recommended it:** Reid Hoffman [^3] - **Key takeaway:** He said the Apple television series contains elements that make him optimistic about the future and highly recommended it [^3] - **Why it matters:** This was Hoffman’s most direct screen recommendation in response to a question about optimism [^3] ### Sherlock - **Content type:** TV series - **Author/creator:** Not specified in the provided notes - **Link/URL:** No direct resource URL was provided; source context: [https://www.youtube.com/watch?v=brscUjmA2To](https://www.youtube.com/watch?v=brscUjmA2To) [^3] - **Who recommended it:** Reid Hoffman [^3] - **Key takeaway:** Hoffman praised Benedict Cumberbatch’s *Sherlock* as an example of taking something old and making it “very present future,” with a through line of humanism [^3] - **Why it matters:** The recommendation points to a specific standard for modernizing older material without losing its human core [^3] > "how you can take this old thing and make it very present future. It's like a through line of humanism." [^3] ## Operator reading and watching ### Winston Churchill’s collected works - **Content type:** Books - **Author/creator:** Winston Churchill [^4] - **Link/URL:** No direct resource URL was provided; source context: [https://x.com/paulg/status/2051979280989511874](https://x.com/paulg/status/2051979280989511874) [^4] - **Who recommended it:** Paul Graham [^4] - **Key takeaway:** Graham highlighted the scale of Churchill’s output and noted that dictation changes the medium, not the cognitive work [^4] - **Why it matters:** The recommendation reads as a standard-setting example of sustained written output while carrying major operating responsibilities [^4] > "He mostly dictated them, but that’s just as much work for the brain, if not for the hand." [^4] ### My first interview with @lulumeservey, Founder of Rostra - **Content type:** Video interview shared on X - **Author/creator:** ti_morse - **Link/URL:** [https://x.com/ti_morse/status/2051740965485113473](https://x.com/ti_morse/status/2051740965485113473) [^5] - **Who recommended it:** Garry Tan [^5] - **Key takeaway:** Tan’s endorsement was simple and strong: he compared Lulu Meservey to “The Wolf” from *Pulp Fiction* [^5] - **Why it matters:** The interview spans topics including Palmer Luckey, Elon, culture after the X acquisition, loyalty, and Jensen Huang’s metric-making, making it a broad operator conversation rather than a narrow founder profile [^6] ## Bottom line If you save one item from today’s set, save **LAB**. It was the only recommendation paired with a clearly stated job to do right now: benchmark what legal agents can actually do and think more concretely about deployment and human-agent collaboration [^1]. --- ### Sources [^1]: [𝕏 post by @saranormous](https://x.com/saranormous/status/2052061665596948894) [^2]: [𝕏 post by @gabepereyra](https://x.com/gabepereyra/status/2052048815998501303) [^3]: [How Beeple's $69 MILLION sale was just the beginning.](https://www.youtube.com/watch?v=brscUjmA2To) [^4]: [𝕏 post by @paulg](https://x.com/paulg/status/2051979280989511874) [^5]: [𝕏 post by @garrytan](https://x.com/garrytan/status/2052007711601291602) [^6]: [𝕏 post by @ti_morse](https://x.com/ti_morse/status/2051740965485113473)