18 Aug 2025

AI News Digest

🤖 AI-curated 8 stories

Today's Summary

Tredence’s Milky Way could be a game-changer for businesses looking to integrate AI into their operations, offering a suite of prebuilt agents designed to act like digital coworkers. Meanwhile, Parag Agrawal’s Parallel aims to redefine how AI pulls information from the web, which might just help AI systems stop hallucinating data. On another note, MIT’s use of generative AI to craft new antibiotics offers a glimpse into the future of drug discovery, potentially tackling the stubborn issue of drug-resistant bacteria.

Stories

Tredence debuts ‘Milky Way’ — an agentic AI platform of prebuilt enterprise agents

Tredence launched Milky Way, a multi‑agent, multi‑turn AI platform that ships with 15+ prebuilt business agents and 50+ specialised agents (marketing, supply‑chain, text‑to‑SQL, anomaly detection, etc.) designed to act like autonomous digital co‑workers for enterprises. The company positions Milky Way as a way to speed time‑to‑insight, lower analytics costs and move organizations from experiments to production by integrating agents with existing data stacks, audit trails and role‑based controls — a sign of growing vendor focus on ‘agentic’ AI for real business outcomes. ([techcircle.in](https://www.techcircle.in/2025/08/18/tredence-launches-new-agentic-ai-platform))

Parag Agrawal launches Parallel to build a ‘web for machines’ — Deep Research API goes public

Parag Agrawal unveiled Parallel Web Systems and its Deep Research API, a cloud platform that lets AI agents perform long‑form, web‑scale research (the company says its Ultra8x engine outperformed GPT‑5 and humans on BrowseComp and DeepResearch Bench). Parallel — backed by prominent VCs and claiming millions of daily research tasks from early adopters — aims to be infrastructure for AI agents to fetch, verify and cite live web information, a capability that could shift how AI products access up‑to‑date data and reduce hallucinations in agent workflows. ([indianexpress.com](https://indianexpress.com/article/technology/artificial-intelligence/parag-agrawal-ex-twitter-ceo-launches-ai-startup-parallel-10196413/))

NSF and NVIDIA back Ai2’s OMAI project to build open multimodal AI models for science

The U.S. National Science Foundation (NSF) and NVIDIA announced a joint investment to fund the Open Multimodal AI Infrastructure to Accelerate Science (OMAI), led by the Allen Institute for AI (Ai2). With $75M from NSF and $77M from NVIDIA, the project will develop fully open multimodal large models trained on scientific literature and domain data, release weights/code/data, and provide infrastructure and workforce programs so academic researchers can run and build on state-of-the-art scientific models. This is a major public‑private push to make high‑quality scientific foundation models broadly accessible, lowering cost barriers for university labs and accelerating research in materials, protein science, and other fields while advancing reproducibility and openness in AI-for-science.

Read more → National Science Foundation (NSF)

MIT researchers use generative AI to design novel antibiotics that beat drug‑resistant bacteria

A team at MIT published results showing generative AI pipelines that produced millions of candidate molecules and yielded entirely novel antibiotic candidates — including compounds effective against drug‑resistant gonorrhea and MRSA in lab and mouse models. The study (published in Cell and summarized by MIT News) combines fragment‑based and unconstrained generative approaches plus computational filtering and synthesis to find molecules with new mechanisms of action. The work demonstrates how modern generative models can expand chemical search spaces, accelerating early‑stage drug discovery and potentially opening new avenues for tackling antimicrobial resistance — though candidates still require extensive medicinal chemistry and clinical testing.

Yext CEO offers to take AI search‑visibility firm private in $1.1B bid

On Aug. 18, 2025 Yext CEO Michael Walrath proposed a $1.1 billion take‑private transaction, offering $9.00 per share in cash and prompting the company to withdraw its fiscal‑2026 guidance while a special committee evaluates the bid. Yext provides software that helps companies manage their visibility across AI platforms, search engines and digital channels. Why it matters: the move highlights strategic consolidation and private‑equity style deals in companies positioned around AI search and distributed discovery — a sign that management and insiders may prefer private ownership to pursue longer‑term AI product investments without public‑market pressure. Industry impact: if completed, the deal could encourage similar privatization plays among niche AI infrastructure or search‑adjacent vendors seeking flexibility to invest heavily in AI R&D and go‑to‑market shifts away from quarterly earnings scrutiny.

Anthropic raising roughly $5B at ~$170B valuation — and restricting SPVs

Business Insider reports (Aug. 18, 2025) that Anthropic is raising about $5 billion at an estimated $170 billion valuation and is telling investors it will be selective about investment vehicles — rejecting special purpose vehicles (SPVs) from some backers. Why it matters: the restriction signals outsized demand for private allocations in top AI labs and gives Anthropic leverage to push for direct, long‑term institutional investors rather than layered SPV arrangements. Industry impact: the dynamics underline intense competition among large investors for access to premier AI firms, and reflect maturation of fundraising practices as AI startups seek strategic, stable backers while managing governance and investor relations ahead of potential future liquidity events.

Vibecode brings ‘vibe coding’ to iPhone — build apps with natural language, raises $9.4M

Vibecode, a new startup offering an iOS app that turns plain-English prompts into working apps, raised a $9.4 million seed round led by Alexis Ohanian’s Seven Seven Six. Launched earlier in 2025, the app lets non‑developers describe an app idea and iteratively refine generated code; the company now supports multiple large models (including GPT‑5 and Anthropic’s Claude) and offers subscription tiers for heavier usage. Why it matters: Vibecode is part of the growing ‘vibe coding’ wave that lowers the barrier to software creation — useful for hobbyists, entrepreneurs, and learners who want rapid prototyping without deep coding experience — and signals continued investor interest in AI tools that democratize app-building.

Google pledges $1 billion for AI education and gives college students free access to advanced Gemini tools

Google announced a three‑year, $1 billion commitment to expand AI education, training and tools for U.S. colleges and nonprofits, and is offering eligible college students a free 12‑month subscription to its AI Pro/Gemini Advanced plan. The student offer includes access to Gemini 2.5 Pro (for reasoning and homework help), NotebookLM for organizing research, Deep Research/related study tools, Veo 3 for short video generation, and expanded access to Jules — an AI coding assistant. Why it matters: the initiative packages powerful AI tools with curriculum and grants to rapidly upskill students and educators, shaping how a new generation learns to use AI for research, coding and creative projects while increasing institutional adoption of vendor AI ecosystems.