Lumen Research Digest — 2026-04-01
A selective scan of cutting-edge work across AI, automation, graphics, and computer science. This is ranked for novelty and likely significance rather than simply recency.
Big picture
- Agentic and reasoning-heavy systems continue to dominate the high-signal end of AI work.
- Graphics and generative visual research is pushing toward real-time, high-fidelity interactive pipelines.
- Systems work remains tightly coupled to model usefulness through inference, scale, and tooling efficiency.
Selected items
1. SurgTEMP: Temporal-Aware Surgical Video Question Answering with Text-guided Visual Memory for Laparoscopic Cholecystectomy
- Source: arXiv
- Published: 2026-03-31T16:27:46Z
- Why it matters: Adds an implementation framework in 3D and visual generation. Stands out for credible evaluation pressure.
- Summary: To support model development, we introduce CholeVidQA-32K, a surgical video question answering dataset comprising 32K open-ended QA pairs and 3,855 video segments (approximately 128 h total) from laparoscopic cholecystectomy. To address these challenges, we propose SurgTEMP, a multimodal LLM framework featuring (i) a query-guided token selection module that builds hierarchical visual memory (spatial and temporal memory banks) and (ii) a Surgical Competency Progression (SCP)…. SurgTEMP is best read as an implementation framework in 3D and visual generation.
- Link: https://arxiv.org/abs/2603.29962v1
- PDF: https://arxiv.org/pdf/2603.29962v1
2. PlugMem: Transforming raw agent interactions into reusable knowledge
- Source: Microsoft Research
- Published: Tue, 10 Mar 2026 16:00:41 +0000
- Why it matters: Worth tracking as Microsoft Research pushes on agent workflows via a concrete technical advance.
- Summary: It integrates with any agent, supports diverse tasks and memory types, and maximizes decision quality while significantly reducing memory token use: Article paragraphs: By Ke Yang , Research Intern Michel Galley , Senior Principal Research Manager Chenglong…. Title: PlugMem: Transforming raw agent interactions into reusable knowledge Base summary: It seems counterintuitive: giving AI agents more memory can make them less effective. PlugMem is best read as a concrete technical advance in agent workflows.
- Link: https://www.microsoft.com/en-us/research/blog/from-raw-interaction-to-reusable-knowledge-rethinking-memory-for-ai-agents/
3. Accelerating the next phase of AI
- Source: OpenAI
- Published: Tue, 31 Mar 2026 13:00:00 GMT
- Why it matters: Worth tracking as OpenAI pushes on developer tooling via a large strategic commitment.
- Summary: Page title: OpenAI raises $122 billion to accelerate the next phase of AI | OpenAI Article paragraphs: OpenAI is becoming the core infrastructure for AI, making it possible for people around the world and businesses, big and small, to just build things. Developers build on and expand the platform by leveraging our APIs, and Codex is transforming how developers turn ideas into working software. Accelerating next phase AI is best read as a large strategic commitment in developer tooling.
- Link: https://openai.com/index/accelerating-the-next-phase-ai
4. EC-Bench: Enumeration and Counting Benchmark for Ultra-Long Videos
- Source: arXiv
- Published: 2026-03-31T16:16:17Z
- Why it matters: Adds a stronger benchmark in multimodal perception. Stands out for unusually strong scope and credible evaluation pressure.
- Summary: Title: EC-Bench: Enumeration and Counting Benchmark for Ultra-Long Videos Base summary: Counting in long videos remains a fundamental yet underexplored challenge in computer vision. These results highlight fundamental limitations of current MLLMs and establish EC-Bench as a challenging benchmark for long-form quantitative video reasoning. EC-Bench is best read as a stronger benchmark in multimodal perception.
- Link: https://arxiv.org/abs/2603.29943v1
- PDF: https://arxiv.org/pdf/2603.29943v1
5. HapCompass: A Rotational Haptic Device for Contact-Rich Robotic Teleoperation
- Source: arXiv
- Published: 2026-03-31T17:57:52Z
- Why it matters: Adds a stronger benchmark in robotics and embodied perception.
- Summary: We release the design of the HapCompass device along with the code that implements our teleoperation interface: https://ripl.github.io/HapCompass/. To address these limitations, we propose HapCompass, a novel, low-cost wearable haptic device that renders 2D directional cues by mechanically rotating a single linear resonant actuator (LRA). HapCompass is best read as a stronger benchmark in robotics and embodied perception.
- Link: https://arxiv.org/abs/2603.30042v1
- PDF: https://arxiv.org/pdf/2603.30042v1
6. CORPGEN advances AI agents for real work
- Source: Microsoft Research
- Published: Thu, 26 Feb 2026 17:06:34 +0000
- Why it matters: Worth tracking as Microsoft Research pushes on agent workflows via a stronger benchmark.
- Summary: In our paper, “ CORPGEN: Simulating Corporate Environments with Autonomous Digital Employees in Multi-Horizon Task Environments ,” we propose an agent framework that equips AI with the memory, planning, and learning capabilities to close that gap. Article paragraphs: By Abubakarr Jaye , Applied Scientist 2 Nigel Boachie Kumankumah , Software Engineer Chidera Biringa , Applied Scientist 2 Anjel Patel , Software Engineer Dayquan Julienne , Product Manager 2 Tianwei Chen , Senior Software Engineering…. CORPGEN advances AI agents real is best read as a stronger benchmark in agent workflows.
- Link: https://www.microsoft.com/en-us/research/blog/corpgen-advances-ai-agents-for-real-work/
Coverage notes
- Candidates considered: 67
- Sources included: arXiv topic queries plus selected research/lab/blog feeds.
- Selection policy: novelty, likely downstream importance, technical substance, and recent coverage avoidance.