Daily enterprise tech brief ·
Enterprise AI releases: 17 November 2026
What changed. Why it matters to your business. What to test.
Lead signals
Research Hub ↗Cohere Adds Video Evidence Retrieval
What's new: Cohere introduced Video Intelligence in Compass private preview, returning timestamped clips, transcripts and visual descriptions from permission-aware enterprise video collections
Teams can make recordings usable as governed agent evidence, reducing time spent locating and validating decisions, disclosures, and operational details in long videos.
Perplexity Launches Monitor API
What's new: Perplexity launched Monitor API for scheduled topic and page monitoring, with relevance filtering, page diffs, and signed webhook delivery through its existing API.
Teams can reduce custom polling and deduplication work for regulatory, competitive, adverse-media, and account-monitoring workflows that trigger downstream systems.
Google Cloud Adds Managed DeepSeek
What's new: Google Cloud added DeepSeek-V4.1-Flash-1017 to Model Garden in Public Preview as a managed multimodal service with up to one million tokens of context.
Platform teams can compare another long-context model inside governed cloud development workflows without operating underlying model infrastructure, subject to preview validation.
Cloud Armor Extends CDN Filtering
What's new: Google Cloud Armor previewed ASN matching, Network Threat Intelligence, and HTTP filtering for globally scoped Cloud CDN edge security policies.
Security teams can test blocking risky network sources and unwanted HTTP traffic at the CDN edge before requests reach origin applications.
Alibaba Releases Qwen4-Max API
What's new: Alibaba released Qwen4-Max through QwenCloud Token Plan and API, offering native vision and a one-million-token context window for complex agentic tasks.
Teams can evaluate a multimodal, long-context API option that may reduce retrieval, chunking, and orchestration steps in selected knowledge and coding workflows.
Tencent Releases Hy4 Flash
What's new: Tencent released Hy4 flash, a multimodal model with a one-million-token context window, through Tencent Cloud TokenHub and several Tencent product surfaces.
Teams can route frequent document, research, and office workflows to a lower-cost model tier while reserving stronger models for complex tasks.
NVIDIA Opens Nemotron for Science
What's new: NVIDIA introduced Nemotron for Science with its Nemotron 3.5 Super VL model, post-training recipes, evaluations, and the nearly 50,000-question LitmusBench v0.2 dataset.
Scientific AI teams can adapt an open model for domain tools and specialized data, supporting more controllable agents for research workflows.
Also on today's Radar
Open any update for the use case, evidence and POC request.
08NVIDIA Expands Vera CPU Platform
What's new: NVIDIA announced a Vera CPU rack-scale platform combining 192 CPUs, Quantum-X800 InfiniBand, and liquid-cooled MGX design; optimized SDK software is available now.
HPC operators can plan integrated CPU, memory, networking, and software capacity for simulation-heavy scientific and AI-assisted workloads arriving in 2027.
09NVIDIA Announces Rubin HPC GPU
What's new: NVIDIA announced Rubin HPC, a future Vera Rubin platform GPU for simulation and data processing, featuring more than 50,000 CUDA cores and seventh-generation NVLink.
Infrastructure leaders can factor accelerated simulation and data movement into long-range HPC and AI-factory plans, although availability is not expected until 2028.
Executive summary & decision signals
POC focus: Teams can make recordings usable as governed agent evidence, reducing time spent locating and validating decisions, disclosures, and operational details in long videos.
- Teams can make recordings usable as governed agent evidence, reducing time spent locating and validating decisions, disclosures, and operational details in long videos.
- Teams can reduce custom polling and deduplication work for regulatory, competitive, adverse-media, and account-monitoring workflows that trigger downstream systems.
- Platform teams can compare another long-context model inside governed cloud development workflows without operating underlying model infrastructure, subject to preview validation.
All source evidence
Practical questions
What enterprise change is described in "Cohere Adds Video Evidence Retrieval"?
Cohere introduced Video Intelligence in Compass private preview, returning timestamped clips, transcripts and visual descriptions from permission-aware enterprise video collections
What enterprise change is described in "Perplexity Launches Monitor API"?
Perplexity launched Monitor API for scheduled topic and page monitoring, with relevance filtering, page diffs, and signed webhook delivery through its existing API.
What enterprise change is described in "Google Cloud Adds Managed DeepSeek"?
Google Cloud added DeepSeek-V4.1-Flash-1017 to Model Garden in Public Preview as a managed multimodal service with up to one million tokens of context.
What enterprise change is described in "Cloud Armor Extends CDN Filtering"?
Google Cloud Armor previewed ASN matching, Network Threat Intelligence, and HTTP filtering for globally scoped Cloud CDN edge security policies.
What enterprise change is described in "Alibaba Releases Qwen4-Max API"?
Alibaba released Qwen4-Max through QwenCloud Token Plan and API, offering native vision and a one-million-token context window for complex agentic tasks.
What enterprise change is described in "Tencent Releases Hy4 Flash"?
Tencent released Hy4 flash, a multimodal model with a one-million-token context window, through Tencent Cloud TokenHub and several Tencent product surfaces.
What enterprise change is described in "NVIDIA Opens Nemotron for Science"?
NVIDIA introduced Nemotron for Science with its Nemotron 3.5 Super VL model, post-training recipes, evaluations, and the nearly 50,000-question LitmusBench v0.2 dataset.
What enterprise change is described in "NVIDIA Expands Vera CPU Platform"?
NVIDIA announced a Vera CPU rack-scale platform combining 192 CPUs, Quantum-X800 InfiniBand, and liquid-cooled MGX design; optimized SDK software is available now.
What enterprise change is described in "NVIDIA Announces Rubin HPC GPU"?
NVIDIA announced Rubin HPC, a future Vera Rubin platform GPU for simulation and data processing, featuring more than 50,000 CUDA cores and seventh-generation NVLink.
Business implications are editorial assessments, not measured returns. Confirm release status, availability and eligibility before testing.
