business-process

Varick Agents 推出面向企业的定制化 AI 智能体服务，端到端自动化部门级运营流程，无需迁移现有系统，通常在 6-12 周内完成从发现到生产部署。

This paper proposes a generative approach using neural temporal point processes for probabilistic forecasting in business process monitoring, demonstrating superior calibration and inference efficiency over existing discriminative models.

Backopsai announces a $42M Series B funding led by Insight Partners to replace human middleware with AI automation, achieving up to 99% resolution in business processes and significant efficiency gains.

The article discusses the distinction between using AI effectively and redesigning work processes to incorporate AI, emphasizing that true automation requires comprehensive process design including triggers, inputs, verification, exception handling, and human decision boundaries.

An AI agency describes how they used Claude to create personalized meme decks for sales outreach, which unexpectedly worked and became their standard outbound process after a reply from a $400M company.

Aaron Levie argues that AI agent adoption will diffuse unevenly across industries because enterprise workflows vary in alignment with continuous digital work, unlike coding where agents thrive. He highlights the need to reengineer business processes for AI agents.

An automation consultant warns that automating workflows without understanding their original purpose can encode outdated rituals and lost context, telling a story of a meaningless 24-hour hold that survived a retired batch system.

This paper proposes an organizational memory architecture for LLM-based agents to execute business processes using shared, organization-specific knowledge, addressing the scalability and consistency challenges of per-agent knowledge encoding.

Most companies mistakenly automate tasks rather than decisions, missing out on significant ROI. By automating decisions that require human judgment, such as lead scoring and support triage, companies can save hours of senior time daily.

A prompt asking which business tasks people would trust an AI agent to handle autonomously versus tasks they would never hand over, reflecting current practical deployment considerations.

An agency founder shares lessons from 50+ AI automation implementations, highlighting that most fail due to broken underlying processes, lack of internal ownership, and over-engineering, while the most successful automations are simple, focused, and backed by a named client-side owner.

The article argues that companies are collections of algorithms and AI will soon optimize every component, leading to a wave of consulting-led transparency and efficiency.

The article critiques ReAct and Workflow agents for enterprise customer service, highlighting issues like lack of controllability and rigidity, then introduces TeliChat, a code-first conversational agent that separates LLM tasks (intent recognition, NLG) from code-driven business logic for better traceability and debuggability.