• 80/20 AI
  • Posts
  • MedChemBench Highlights AI Progress in Drug Discovery

MedChemBench Highlights AI Progress in Drug Discovery

If Claude Can Help Prove the Riemann Hypothesis, What Would You Tackle With It?

Sponsored by

The Agentic Engineering Playbook top engineers are using

Most engineers are stuck babysitting AI one prompt at a time. A handful have figured out how to build agents that plan, execute, and self-correct on their own. Sign up for The Code and get The Complete Playbook for Agentic Engineering. Claude Code, subagents, MCP, all of it.

MedChemBench Highlights AI Progress in Drug Discovery

Your AI Agent Can Now Spend Your AWS Money

AWS Marketplace now lets AI agents find and buy select products (APIs, data, software) and put the charge straight on your existing AWS bill, in US dollars. It runs through Amazon Bedrock AgentCore payments using the x402 agent payment protocol, so there's no crypto wallet or separate API key to set up.

The guardrails exist, but you have to configure them. You set spending limits before an agent can transact, Private Marketplace restricts which products agents can buy, and every purchase shows up in CloudTrail and Cost Explorer. An agent's first purchase from a seller creates an agreement with your account, and later purchases are tracked against it automatically.

Before you switch it on:

  • Set a per-transaction and a total spending limit in AgentCore payments first.

  • Turn on Private Marketplace so agents can only buy from an approved list.

  • Add a budget alert for Marketplace charges in Cost Explorer, and check CloudTrail weekly.

Why it matters: an agent that can spend is like a new hire with a company card. Give it a limit before you give it a task.

AI NEWS HIGHLIGHT

If Claude Can Help Prove the Riemann Hypothesis, What Would You Tackle With It?

Where it stands: Alpöge released the proof as Lean code on Nov 20. Independent checkers compiled it without errors, and the Lean community says it proves the exact Mathlib statement of the hypothesis. Wider expert review is still under way.

Levent Alpöge directed the architecture; Claude handled the hardest algebraic work. The result — if verified — is the most complex thing any human–AI collaboration has ever produced. The pattern is now clear: AI doesn't replace the expert; it removes the ceiling on what that expert can reach. Use these prompts to apply that same pattern to the hardest unsolved problems in your own domain.

Ask your AI

1. Alpöge's approach worked because the Riemann Hypothesis has a precise formal structure — it can be stated in Lean 4, checked mechanically, and built incrementally. What is the equivalent "precisely formalizable" problem in my field — [product analytics / pricing optimization / API design / security architecture]? Walk me through how to frame it as something an AI can help solve systematically rather than just advise on.

2. The Riemann proof's biggest verification challenge is not correctness — the Lean kernel confirmed that — but whether the formal theorem statement actually matches the classical hypothesis. In my AI-assisted work, what is the equivalent gap between "what the AI produced" and "what I actually needed"? Help me design a verification protocol for AI-generated work products that catches this category of error before I act on them.

3. Replit's study showed 32 Claude Opus agents achieved an 18.9x speedup on optimization tasks — but GPT-6.2 Astra degraded with more agents. I want to run a multi-agent workflow for [describe your task]. Help me design the agent team: how many agents, what roles, what coordination structure, and how to detect when adding more agents is hurting rather than helping.

End with: a one-page "human-AI collaboration design" brief — the problem, the approach, where the AI leads, where I verify, and how I'll know it worked.

TOP TRENDING AI TOOLS

  • Mistral Large 4 — 1T-param multimodal MoE, 52B active; API preview at $1.36/$4.18 per million tokens; open weights coming end of October.

  • Recursive Flint 39.8B MoE — 2.5B active, 256 experts, 262K context, Apache 2.0; runs ~125 tokens/sec on M4 Pro; 89.7% SWE-bench Pro v2 (self-reported).

  • Claude Duet — Proactive daily-use agent app (Opus 5.5) with morning briefs, cloud computer, and 6,000+ integrations; US Pro/Max waitlist.

  • Walmart Agent Gateway (MCP) — MCP server for approved AI agents to search inventory and purchase on customers' behalf with built-in spending controls.

  • Notion Agent Harness — 24+ hour single-task agent runs with Auto router claiming ~90% cost reduction versus direct model calls.

  • Hugging Face Carbon-X — 14B genomic LLM, 300B training tokens, 50K base pairs per query, Apache 2.0; no wet-lab validation yet.

  • WorldTUI (Simon Edwardsson) — 360M-parameter model predicts terminal screen changes per keystroke; completes 80/106 ShellBench commands; p95 latency 0.87s on RTX 4090.

A CRM so smart, it updates itself.

HubSpot's CRM is so smart, it now updates itself. Calls get logged and summarized the moment they end. New leads get researched and a first outreach drafted, automatically. Deals move forward on their own, with next steps flagged before you have to ask. 

Even the one repetitive task you've been meaning to fix can now run itself, no code required. Customers using it are already seeing the difference. 

This isn't the CRM you have to work hard to figure out. It's the one already working before you log in.

SPONSOR US

Get your business in front of over 90k+ AI professionals

8020AI is the world’s #1 AI Newsletter, Read by 90k+ professionals from leading companies such as Google, OpenAI, Meta, and Microsoft.

We've assisted in promoting Over 500 AI-Related Products. Will yours be the next?

What We Can Offer:

  • Launch an Advertising Campaign

  • Introduce New Product or Features

  • Other Business Cooperation

Or Email our founder Alamin at [email protected]

FEEDBACK

How was your experience with 8020AI today?

How was 8020AI today?

Login or Subscribe to participate in polls.

Login or Subscribe to participate in polls.

If you have specific feedback or anything interesting you’d like to share, please let us know by replying to this email.