Intuit’s outcome-grounded LLM pattern for financial advice

#1 𝕏

Udi Menkes 🚢 shared AI Engineer’s video of his talk on Intuit’s financial-advice system, which derives millions of business state–action–outcome trajectories, uses reinforcement learning to select better actions, and trains an LLM to generate advice. Intuit reported that a cheaper midsize model grounded in this data outperformed leading models.

#2 𝕏

Garry Tan described YC’s QM as a free, open-source multi-agent harness for personal AI or company-wide use across accounting, legal, events, and engineering. Available under the MIT license with Slack and web interfaces, QM is used daily by Tan’s team.

#3 in

Guillermo Rauch announced advanced spend budgets for Vercel AI Gateway, including budgets per key, team, and project. Vercel’s AI Gateway also offers failover, model and provider choice, and realtime observability.

#4 𝕏

Aravind Srinivas called the DeepSeek V4-Flash performance/cost improvements “a big deal,” noting that two-orders-of-magnitude improvements are rare.

#5 𝕏

Guillermo Rauch shared an open-source agentic CRM built on eve.dev and Next.js. He described it as model-agnostic, headless, multi-channel, and deployable via self-hosting or serverlessly.

#6 𝕏

Garry Tan characterized OpenAI’s most interesting 2026 “vibe shift” as appearing to become an open platform, contrasting intelligence offered as a utility with signals that full-stack integration is optimal. His post quoted a discussion comparing Anthropic’s strategy with OpenAI’s.

#7 𝕏

Lenny Rachitsky commented that “everyone is becoming a part-time engineer and marketer,” responding to Ethan Mollick’s post on AI blurring job lines and OpenAI findings about porous organizational boundaries.

#8 𝕏

Shreyas Doshi commented “Taste” on Andrew Chen’s post about unlimited AI-generated content versus limited human capacity to verify proofs, code, videos, and more.

#9 in

Peter Yang said he thinks Opus 4.6 had the best personality and writing style among Opus models, while Opus 5 tends to give him overly long replies, uses too much “Claude-speak,” and is too judgmental. He said Opus once felt like a trusted friend, but no longer does.