Nexla ran its Inspiration to Trip demonstration on Nebius AI Cloud at NVIDIA GTC in March 2026. Built with NVIDIA and Tripadvisor, the pipeline analyzes travel-video frames and audio, enriches extracted locations and experiences with travel data, and generates personalized itineraries. Users submit a video link through an experimental feature in Tripadvisor’s native AI experience. Enrichment adds live ratings, images, availability and personalization signals. Conversational refinement works without rerunning the pipeline. Nexla describes dedicated, non-virtualized NVIDIA HGX B200 systems with Quantum InfiniBand networking and API access to NVIDIA models provisioned through Nebius.
Nebius also describes Nexla as running its multi-agent pipelines on Nebius AI Cloud inference infrastructure. The demonstration uses NVIDIA Nemotron Nano 12B VLM for visual analysis, NVIDIA ASR models for transcription and Nemotron 3 Super for itinerary generation. In its architecture account, Nexla also describes OpenAI Whisper transcription running in parallel with frame analysis, with synchronized outputs passed to summarization. Frame sampling is configurable at 3, 5, 10 or 15 frames per second. Visual analysis extracts landmarks, geographic markers, on-screen text, objects, people and entities. Nexla reports more accurate itineraries and richer recommendations than general-purpose foundation models in the same workflow. It also reports reduced prompt-engineering and orchestration effort with NVIDIA Nemotron models.
Nexla reports up to 85% cost efficiency for NVIDIA NIM microservices running on Nebius compared with leading cloud-hosted API models, with comparable output quality. Its benchmarks were preliminary tests on sample videos, projected to a million-video production scenario.
Nexla says the NVIDIA models in its travel-video pipeline are provisioned through Nebius for API access. Videos are sampled at 3, 5, 10 or 15 frames per second. Nemotron VL Nano 12B v2 extracts landmarks, geographic markers, on-screen text, objects, people and entities. OpenAI Whisper transcription runs in parallel with frame analysis, and synchronized outputs feed itinerary generation.
Video-to-itinerary pipeline architecture16 Mar 2026 · Nexla · Web pageNexla says the NVIDIA models in its travel-video pipeline are provisioned through Nebius for API access. Videos are sampled at 3, 5, 10 or 15 frames per second. Nemotron VL Nano 12B v2 extracts landmarks, geographic markers, on-screen text, objects, people and entities. OpenAI Whisper transcription runs in parallel with frame analysis, and synchronized outputs feed itinerary generation.
GTC travel-planning pipeline on B200
16 Mar 2026 · Nexla · Web page
Nexla says its Inspiration to Trip demonstration runs on Nebius AI Cloud using dedicated NVIDIA HGX B200 systems with Quantum InfiniBand networking. The workflow analyzes travel videos, enriches extracted information with travel data and generates personalized itineraries with conversational refinement. Nexla reports up to 85% cost efficiency for NVIDIA NIM microservices on Nebius compared with leading cloud-hosted API models, with comparable output quality.
GTC travel-planning pipeline on B20016 Mar 2026 · Nexla · Web pageNexla says its Inspiration to Trip demonstration runs on Nebius AI Cloud using dedicated NVIDIA HGX B200 systems with Quantum InfiniBand networking. The workflow analyzes travel videos, enriches extracted information with travel data and generates personalized itineraries with conversational refinement. Nexla reports up to 85% cost efficiency for NVIDIA NIM microservices on Nebius compared with leading cloud-hosted API models, with comparable output quality.
Nebius Nexla pipelines on Nebius AI Cloud
27 Sept 2026 · Nebius · Web page
Nebius says Nexla runs its multi-agent pipelines on Nebius AI Cloud inference infrastructure. It describes the partnership’s GTC 2026 debut through a travel-planning system built with Tripadvisor.
Nebius Nexla pipelines on Nebius AI Cloud27 Sept 2026 · Nebius · Web pageNebius says Nexla runs its multi-agent pipelines on Nebius AI Cloud inference infrastructure. It describes the partnership’s GTC 2026 debut through a travel-planning system built with Tripadvisor.
From fragmented data to production-grade agents: Nebius, Nexla, and Tripadvisor at NVIDIA GTC
17 Mar 2026 · nebius.com · Web page
Nebius describes a joint stack in which Nexla orchestrates data and agent workflows while Nebius supplies dedicated NVIDIA infrastructure. Nexla, Nebius, NVIDIA and Tripadvisor demonstrated the architecture through Inspiration to Trip at GTC.
From fragmented data to production-grade agents: Nebius, Nexla, and Tripadvisor at NVIDIA GTC17 Mar 2026 · nebius.com · Web pageNebius describes a joint stack in which Nexla orchestrates data and agent workflows while Nebius supplies dedicated NVIDIA infrastructure. Nexla, Nebius, NVIDIA and Tripadvisor demonstrated the architecture through Inspiration to Trip at GTC.
Accelerate GenAI RAG with Nexla and NVIDIA
15 Oct 2026 · nexla.com · Web page
Nexla describes a travel-video itinerary demonstration using NVIDIA AI, Tripadvisor data, Nebius infrastructure and Nexla’s orchestration platform. It reports more accurate itineraries, richer recommendations and reduced prompt-engineering and orchestration effort with NVIDIA Nemotron models.
Accelerate GenAI RAG with Nexla and NVIDIA15 Oct 2026 · nexla.com · Web pageNexla describes a travel-video itinerary demonstration using NVIDIA AI, Tripadvisor data, Nebius infrastructure and Nexla’s orchestration platform. It reports more accurate itineraries, richer recommendations and reduced prompt-engineering and orchestration effort with NVIDIA Nemotron models.
Nexla makes enterprise software that connects, transforms, and delivers data for AI agents, analytics, and business operations. Its platform supports databases, applications, files, documents, and streaming sources, turning data into reusable products called Nexsets. Main offerings include MCP Studio for building governed connections between agents and enterprise systems, Agentic RAG for retrieval workflows, Express.Dev for creating pipelines through prompts, and Helix for supplying data context. Enterprise customers listed by Nexla include DoorDash, LinkedIn, LiveRamp, and Johnson & Johnson.
Nexla outlines enterprise connectivity for Snowflake Cortex Agents
Nexla announced support for connecting Snowflake Cortex Agents and CoWork to enterprise systems through task-specific MCP servers. The integration provides governed access to live data and write actions, alongside data pipelines and zero-copy access across more than 1,000 systems.
Data integration and orchestration for AI agents · Infrastructure
How they use Nebius AI Cloud
Nexla ran its Inspiration to Trip demonstration on Nebius AI Cloud at NVIDIA GTC in March 2026. Built with NVIDIA and Tripadvisor, the pipeline analyzes travel-video frames and audio, enriches extracted locations and experiences with travel data, and generates personalized itineraries. Users submit a video link through an experimental feature in Tripadvisor’s native AI experience. Enrichment adds live ratings, images, availability and personalization signals. Conversational refinement works without rerunning the pipeline. Nexla describes dedicated, non-virtualized NVIDIA HGX B200 systems with Quantum InfiniBand networking and API access to NVIDIA models provisioned through Nebius.
Nebius also describes Nexla as running its multi-agent pipelines on Nebius AI Cloud inference infrastructure. The demonstration uses NVIDIA Nemotron Nano 12B VLM for visual analysis, NVIDIA ASR models for transcription and Nemotron 3 Super for itinerary generation. In its architecture account, Nexla also describes OpenAI Whisper transcription running in parallel with frame analysis, with synchronized outputs passed to summarization. Frame sampling is configurable at 3, 5, 10 or 15 frames per second. Visual analysis extracts landmarks, geographic markers, on-screen text, objects, people and entities. Nexla reports more accurate itineraries and richer recommendations than general-purpose foundation models in the same workflow. It also reports reduced prompt-engineering and orchestration effort with NVIDIA Nemotron models.
Nexla reports up to 85% cost efficiency for NVIDIA NIM microservices running on Nebius compared with leading cloud-hosted API models, with comparable output quality. Its benchmarks were preliminary tests on sample videos, projected to a million-video production scenario.
Nexla says the NVIDIA models in its travel-video pipeline are provisioned through Nebius for API access. Videos are sampled at 3, 5, 10 or 15 frames per second. Nemotron VL Nano 12B v2 extracts landmarks, geographic markers, on-screen text, objects, people and entities. OpenAI Whisper transcription runs in parallel with frame analysis, and synchronized outputs feed itinerary generation.
Video-to-itinerary pipeline architecture16 Mar 2026 · Nexla · Web pageNexla says the NVIDIA models in its travel-video pipeline are provisioned through Nebius for API access. Videos are sampled at 3, 5, 10 or 15 frames per second. Nemotron VL Nano 12B v2 extracts landmarks, geographic markers, on-screen text, objects, people and entities. OpenAI Whisper transcription runs in parallel with frame analysis, and synchronized outputs feed itinerary generation.
GTC travel-planning pipeline on B200
16 Mar 2026 · Nexla · Web page
Nexla says its Inspiration to Trip demonstration runs on Nebius AI Cloud using dedicated NVIDIA HGX B200 systems with Quantum InfiniBand networking. The workflow analyzes travel videos, enriches extracted information with travel data and generates personalized itineraries with conversational refinement. Nexla reports up to 85% cost efficiency for NVIDIA NIM microservices on Nebius compared with leading cloud-hosted API models, with comparable output quality.
GTC travel-planning pipeline on B20016 Mar 2026 · Nexla · Web pageNexla says its Inspiration to Trip demonstration runs on Nebius AI Cloud using dedicated NVIDIA HGX B200 systems with Quantum InfiniBand networking. The workflow analyzes travel videos, enriches extracted information with travel data and generates personalized itineraries with conversational refinement. Nexla reports up to 85% cost efficiency for NVIDIA NIM microservices on Nebius compared with leading cloud-hosted API models, with comparable output quality.
Nebius Nexla pipelines on Nebius AI Cloud
27 Sept 2026 · Nebius · Web page
Nebius says Nexla runs its multi-agent pipelines on Nebius AI Cloud inference infrastructure. It describes the partnership’s GTC 2026 debut through a travel-planning system built with Tripadvisor.
Nebius Nexla pipelines on Nebius AI Cloud27 Sept 2026 · Nebius · Web pageNebius says Nexla runs its multi-agent pipelines on Nebius AI Cloud inference infrastructure. It describes the partnership’s GTC 2026 debut through a travel-planning system built with Tripadvisor.
From fragmented data to production-grade agents: Nebius, Nexla, and Tripadvisor at NVIDIA GTC
17 Mar 2026 · nebius.com · Web page
Nebius describes a joint stack in which Nexla orchestrates data and agent workflows while Nebius supplies dedicated NVIDIA infrastructure. Nexla, Nebius, NVIDIA and Tripadvisor demonstrated the architecture through Inspiration to Trip at GTC.
From fragmented data to production-grade agents: Nebius, Nexla, and Tripadvisor at NVIDIA GTC17 Mar 2026 · nebius.com · Web pageNebius describes a joint stack in which Nexla orchestrates data and agent workflows while Nebius supplies dedicated NVIDIA infrastructure. Nexla, Nebius, NVIDIA and Tripadvisor demonstrated the architecture through Inspiration to Trip at GTC.
Accelerate GenAI RAG with Nexla and NVIDIA
15 Oct 2026 · nexla.com · Web page
Nexla describes a travel-video itinerary demonstration using NVIDIA AI, Tripadvisor data, Nebius infrastructure and Nexla’s orchestration platform. It reports more accurate itineraries, richer recommendations and reduced prompt-engineering and orchestration effort with NVIDIA Nemotron models.
Accelerate GenAI RAG with Nexla and NVIDIA15 Oct 2026 · nexla.com · Web pageNexla describes a travel-video itinerary demonstration using NVIDIA AI, Tripadvisor data, Nebius infrastructure and Nexla’s orchestration platform. It reports more accurate itineraries, richer recommendations and reduced prompt-engineering and orchestration effort with NVIDIA Nemotron models.
Nexla makes enterprise software that connects, transforms, and delivers data for AI agents, analytics, and business operations. Its platform supports databases, applications, files, documents, and streaming sources, turning data into reusable products called Nexsets. Main offerings include MCP Studio for building governed connections between agents and enterprise systems, Agentic RAG for retrieval workflows, Express.Dev for creating pipelines through prompts, and Helix for supplying data context. Enterprise customers listed by Nexla include DoorDash, LinkedIn, LiveRamp, and Johnson & Johnson.
Nexla outlines enterprise connectivity for Snowflake Cortex Agents
Nexla announced support for connecting Snowflake Cortex Agents and CoWork to enterprise systems through task-specific MCP servers. The integration provides governed access to live data and write actions, alongside data pipelines and zero-copy access across more than 1,000 systems.
