Genspark.aiTranslation site

Company Overview

Genspark.ai, commonly referred to as Genspark, is a startup and product brand centered on AI workspaces and general-purpose agents. According to investor materials, the company was founded in 2023. Its co-founders include Eric Jing and Kay Zhu. Jing serves as CEO, while Zhu serves as CTO; the team has a background in search and AI product development.

Genspark has a close relationship with MainFunc: MainFunc’s official website lists Genspark as one of its core supported products, while Genspark’s privacy policy lists MainFunc and its subsidiaries or affiliated entities—including Genspark Inc.—collectively as service providers. Therefore, MainFunc should not be simply regarded as the product’s former name. The product is designed for individual knowledge workers and teams, with a focus on transforming requirements into deliverables that can be continuously edited and utilized.

Milestones

On June 18, 2024, Genspark publicly launched Sparkpages, an AI-powered search tool that organizes information from various sources into themed pages. On April 2, 2025, the company introduced Super Agent, expanding its capabilities to include task planning, tool integration, and the generation of a variety of outputs.

On July 20, 2026, AI Workspace 6.0 was released, with memory, agents, toolkits, and collaboration as its core layers. On September 10 of the same year, the company released Gen-1 Slides, which enhanced AI Slides with post-trained models tailored for presentation tasks. From search entry points to workspaces and then to specialized models, the focus of development has gradually shifted toward the quality of execution for complete tasks.

Products & Services

• Super Agent:Plan and execute multi-step tasks based on user objectives, combining models, tools, and data sources to support research, content creation, and other knowledge workflows.
• Office Suite:These include Slides, Sheets, Docs, and Gmail, which are used for presentations, spreadsheets, documents, and email-related tasks; users can adjust the content using subsequent commands.
• Build Suite and Content Suite:The former provides tools such as design, coding, and AgentBase to support the development of websites, dashboards, and internal systems; the latter covers content creation such as conversations, images, videos, and music.
• SecondBrain:Organize users' emails, meetings, documents, and project materials into a continuous context to reduce the need to repeatedly provide background information when performing related tasks.
• GenTeam:Enable members and AI agents to work together in a shared collaborative space, supporting the deployment of agents or the creation of task assistants as needed.

core technology

Genspark emphasizes collaboration among multiple models and agents. The system invokes the appropriate models and tools based on different tasks, then integrates information retrieval, data processing, and output generation. The challenge of this architecture lies not only in generating a response but also in managing context, selecting tools, handling exceptions during execution, and ensuring consistency across the results of multiple steps.

Gen-1 Slides, on the other hand, demonstrates a task-specific optimization approach: According to the official description, this model is based on the open-weight MiniMax M3 and was trained in collaboration with Fireworks AI for use in AI Slides’ Standard mode. The platform thus employs both external model orchestration and task-specific fine-tuning. Generated figures, citations, and formatting still require verification, and when linking to personal or corporate data, the scope of data access and usage should also be verified.

For example, creating an industry report typically involves first gathering data, then extracting key metrics, and finally organizing charts and pages. When the workspace organizes these steps into a sequence of tasks, users can continue to request revisions based on the results without having to redefine their objectives in each tool. However, data collection, metric calculation, and page generation are distinct quality control stages, and an error in any one step could affect the final deliverable.

Sustained memory also requires clear boundaries: which data is used for the current project, which can be reused across tasks, and whether old information is still valid—all of these factors influence the results. When implementing this approach, teams should prioritize linking essential data while retaining the ability to verify context and outputs.

development prospect

Given the expansion of its integrated office suite and team collaboration capabilities, Genspark is expected to reduce the burden on users of switching between search, analysis, writing, and formatting tools. If continuous context reliably supports subsequent tasks, the platform may foster consistent usage habits in repetitive reporting, project documentation, and daily content creation.

This outlook still depends on the success rate of long-term task execution, the editability of results, the cost of invoking the model, and permission management. The number of features alone does not guarantee delivery quality; users are better off evaluating, through real-world tasks, whether the results are accurate, whether they are easy to modify, and how much manual rework is required before deciding which tasks to delegate to the agent on a long-term basis.