The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Back to feedPost
Anthropic reports that about 950 Claude agents searched DNA data for 21 hours, used 210M tokens, and surfaced a previously uncharacterized enzyme system. The useful pattern: agents generate candidates at scale, then experts review and test them in the lab. #AI
The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Related posts
What happens when AI pulls you into a field you know nothing about
I started this project knowing nothing about chemistry. I finished it reading molecular structures.
VALENCE is a speculative fragrance brand, but really it was an experiment: can AI help you go deeper into a concept, not just faster to a pretty image?
Every perfume is a formula, a handful of real molecules with real shapes. So instead of inventing a "scientific look," I built the whole brand from the actual chemistry. Three scents, nine molecules, all verified in PubChem. Geosmin, the molecule behind the smell of rain on dry earth, gave the scent Petrichor its name.
The process went one way only: from real data to image, never the other way round. Molecular geometry calculated in RDKit, molecules rendered as glass sculptures with a Python path tracer built together with Claude, and those renders used as references for AI-generated campaign visuals, motion and packaging. The bottle went through many iterations before it landed on one calm form: clear glass, a chrome sphere, each scent told apart only by its colour.
What I took away: an AI workflow can connect design with fields that seem far from it, and turn that connection into a campaign.
Tools: RDKit, Python, Claude, Midjourney, Flora, Kive
Amazing!!
Local developer tool that executes atomic commit plans through a dual-agent AI pipeline (Generator & Reviewer) with real-time SSE streaming, quality gates, and automated rollbacks.
The dual agent pipeline creating atomic commit plans and the live SSE stream makes the review loop feel almost instant.
Completed "Building and Evaluating Advanced RAG" by DeepLearning.AI (TruEra, LlamaIndex).
The most useful part wasn't building another RAG pipeline, it was learning how to actually evaluate one. The RAG Triad (answer relevance, context relevance, groundedness) gives you a way to catch exactly where a system is failing: bad retrieval vs. the model making things up vs. just answering the wrong question.
Also went through sentence-window and auto-merging retrieval, two ways to fix the classic RAG trade-off between precise matching and having enough context to actually answer well.
Applying this directly to a RAG tool I'm building for equity research. Evaluation is the part of RAG most people skip, and it's exactly where things quietly break in production.
https://www.deeplearning.ai/accomplishments/491d7a76-f89d-4e13-9d6b-3377fced78a2?accomplishmentId=491d7a76-f89d-4e13-9d6b-3377fced78a2&usp=sharing
evaluating RAG is the part nobody talks about, the triad is super handy