Skip to content

AI Enterprise & Model Brief

Go.AI Raises $85M Series A as OpenAI Tests Wet-Lab Biology AI

A 16,000-site SEOcrawl study finds AI referrals up 45% year-over-year, while Physicl raises $10M, Underdog ships a 32MB voice model, and Airbnb plans a 2027 AI travel agent.

By Lazy Tensor
Published · 5 min read

Enterprise AI platforms, wet-lab biology benchmarks, and on-device models reshape deployment economics in November 2026.
Enterprise AI platforms, wet-lab biology benchmarks, and on-device models reshape deployment economics in November 2026. · Lazy Tensor / Editorial Visual Desk
Your listening Queue & catch up
Unheard stories first, timed at 1×. Your selections stay first.

Add a story or build a catch-up queue to start listening.

Enterprise artificial intelligence investment, empirical traffic measurement, and scientific benchmarking converged this week across six major industry disclosures spanning conversational agents, search referral dynamics, wet-lab biology evaluation, physical simulation, edge speech synthesis, and consumer travel.

Go.AI Secures $85 Million Series A Led by Updata Partners

Conversational AI platform Go.AI announced on November 7, 2026 that it has closed an $85 million Series A funding round led by growth equity firm Updata Partners, bringing the company's total capital raised to $90 million. Founded in 2024 by domain broker and investor Andrew Rosener, Go.AI operates from its flagship Go.com domain to automate customer acquisition, lead qualification, and support workflows across voice, SMS, email, and web chat.

Over the past 12 months, Go.AI expanded to more than 200 enterprise customers across financial services, healthcare, telecommunications, e-commerce, and home services, achieving an 8x year-over-year increase in annual recurring revenue (ARR). Its autonomous agents currently handle over 12.5 million daily conversational queries across more than 45 languages and integrate natively with Salesforce, HubSpot, Zendesk, and ServiceNow. Early customer Braden Pollock, Founder of Legal Brand Marketing, noted that deploying Go.AI improved speed-to-lead and conversion metrics while reducing operational overhead. As part of the financing, Updata Partners General Partner Jon Seeber and Partner Carter Griffin joined the Go.AI Board of Directors as the company expands across North America, Europe, and Asia-Pacific.

SEOcrawl Study: AI Visits Grow 45% Across 3,000+ Websites but Trail Google 80-to-1

While enterprise agent adoption accelerates, a 12-month global study published by SEO analytics platform SEOcrawl on November 7, 2026 provides a sober empirical baseline for AI referral traffic. Analyzing anonymized traffic data from October 2025 through October 2026 across more than 16,000 websites—including an active AI cohort of more than 3,000 websites that record at least 1,000 monthly visits and at least 1 AI referral—researchers found that AI-referred sessions grew 45% year-over-year. Despite that growth, AI referrals increased from 0.24% of total website traffic to 0.36%, remaining below 1% of overall web visits.

Across the dataset, websites received an average of 1.25 AI visits for every 100 visits generated by Google Search, giving traditional search an 80x volume lead. However, organic clicks from Google Search declined 15% year-over-year even as search impressions held steady, reflecting the expansion of AI Overviews and zero-click search answers. Among AI referral sources, OpenAI's ChatGPT accounted for 83% of all AI-referred visits, followed by Google Gemini at 6%, Perplexity at 5%, Anthropic's Claude at 4%, and Microsoft Copilot at 2%. SEOcrawl CEO and Founder Lokesh emphasized that brands must track AI visibility alongside traditional search optimization rather than replacing core SEO strategies.

OpenAI and Ligo Release BioBasicsBench for Wet-Lab Biology

To evaluate whether frontier language models understand practical laboratory execution rather than textbook trivia, OpenAI and biological AI company Ligo introduced BioBasicsBench, an open benchmark of 100 multiple-choice questions (each offering 4 options with 1 correct answer) derived from more than 200 real-world experimental contexts. Every item required agreement from at least 2 graduate-level wet-lab biologists and an 80% consensus threshold in blind review across 5 core categories: Experimental Design and Hypothesis Formulation, Protocol Execution and Parameter Choice, Experimental Artifact and Failure Mode Identification, Data and Readout Interpretation, and Quantitative Lab Math (including molarity, serial dilutions, and unit conversions).

Questions test hands-on failure modes—such as whether membrane-impermeant SYTOX nucleic acid dyes used to assess cell viability remain stably retained after sample fixation. OpenAI evaluated 7 frontier models across 37 model-effort settings, executing 5 independent runs per question for a total of 18,500 test results. GPT-6 Astra achieved the highest accuracy on the full benchmark at 66.7% under both high and xhigh reasoning effort—rising 17.7 percentage points from 49.0% at none effort—while GPT-6.1 Sol scored 62.0% and GPT-5.6 Sol reached 43.8% at xhigh effort. OpenAI released the dataset on Hugging Face to support reproducible biosecurity and scientific evaluation.

Physicl Raises $10M for Physical AI Simulation; Underdog Ships 32MB On-Device Speech Model

In robotics and edge computing, physical AI simulation startup Physicl raised a $10 million seed round led by Inovia Capital with participation from angel investors at Nvidia Corp. and Alphabet Inc.'s DeepMind. Founded by CEO Alex de Vigan—who previously founded 3D product visualization company Nfinite, serving Lowe's, Staples, and Wayfair—Physicl provides its Physicl Studio library of 10,000 indoor and outdoor 3D environments and 100,000 simulated 3D objects, paired with partner SpAItial's Echo-3 world foundation model, to stress-test warehouse robots and autonomous vehicles using synthetic camera and lidar data.

On the consumer edge, open-source AI lab Underdog, founded by Sigil Wen, released two Apache 2.0 audio models engineered to run locally on standard CPUs without cloud servers or dedicated GPUs: Woof, a 32-megabyte (31 million parameter) text-to-speech model capable of cloning a voice from a 5-second audio sample, and Bark, a 167-megabyte speech-to-text transcription engine, alongside a 27-billion-parameter general intelligence model running locally on a consumer laptop. Rounding out consumer deployment plans, Airbnb CEO Brian Chesky told investors during Thursday's Q3 2026 earnings call that the company will launch an interactive AI travel agent in the first half of 2027—building on internal research showing 80% of travelers enjoy researching trips, active AI customer service pilots in the U.S., Mexico, and Canada, and Q3 2026 revenue that rose 19% year-over-year to $4.87 billion.

Sources

Follow the coverage

Published reporting connected by Venture Capital.

  1. Anthropic S-1 Details $16.3B Revenue as Deepgram Buys Daily & Wabi Up
  2. Sarah Guo and the Price of Conviction
  3. Elad Gil and the Economics of Being Early
  4. Go.AI Raises $85M Series A as OpenAI Tests Wet-Lab Biology AIYou are reading this story

Comments (0)

No comments yet. Be the first to share a thought on this story.

The Lazy Tensor briefing

Get the next story in your inbox.

Independent reporting on the people, companies and ideas shaping AI and technology. Free, once a week.

More from Lazy Tensor