Talgent
Talgent builds an enterprise AI-agent platform intended to move organizations from individual AI productivity to coordinated, AI-native work. Talgent Workbench gives each person or registered program an agent in an isolated workspace, while Talgent Orchestra connects agents across projects so changes reach affected people and completed work becomes organizational knowledge.
At a Glance
- Enterprises and organizations seeking governed AI agents
- Teams with cross-functional projects and coordination overhead
- Organizations requiring self-hosting, auditability and control of AI data and telemetry
- Developers and product teams embedding agent workflows through public APIs
- +1 more
AI Tools by Talgent
(1)Talgent Workbench
Self Hosted AI Agent Platform
Discussions
No discussions yet
Be the first to start a discussion about Talgent
Latest News
Workbench release: a Work fixes its Work Caller, Work Key and File Home at creation (commit #53).
Self-hosting guide updated for the Kubernetes deployment architecture and prerequisites.
Applications can drive Workbench through its public APIs, including starting Works and collecting Artifacts and usage.
The v1 PersonalWorkService was removed in favor of agent.public.v2.WorkService.
Products & Services
Open-source, self-hosted managed-agent platform. Each person or registered program gets an AI Agent; a Work can receive instructions, files, repositories, a selected model and Plugins, then run in its own sandbox or on a connected computer.
A product built on Workbench that adds project Intents/goals with owners, a coordinator that routes changes to affected agents, decision traceability and project knowledge derived from finished work.
A managed-computer client/package that connects a user’s Mac, Linux or Windows machine to Workbench so a Work can run with the machine’s files, tools and apps; the README directs users to install @talgent/computer from npm.
Public APIs and OAuth 2.0 client-credential flow for registered applications to start Works, follow them, collect Artifacts and inspect usage; applications receive their own Agent, Plugins and authorizations.
Market Position
Talgent positions Workbench as an open-source, self-hosted alternative to sending company work and know-how to a hosted AI provider: the control plane, agent conversations, files and telemetry can run in the customer’s Kubernetes cluster, with model choice per Work. Its differentiation is the combination of a personal agent workbench, local/cloud execution, programmatic Applications and Orchestra’s cross-agent coordination and organizational-memory layer.
Leadership
Founders
Harry Hsu
The public @harryhsu88 X profile describes him as ‘building Talgent.’ No prior roles or companies were identified in the reviewed sources.
jacexh
GitHub user jacexh is the principal visible maintainer/author of the Talgent Workbench repository and posted the project’s Show HN launch. The public GitHub profile lists Suzhou, China and earlier projects including a Go HTTP load-testing tool, a Python AutoItX3 binding, and a Go DDD template; no legal name was provided.
Founding Story
The project is presented as a response to the gap between faster individual AI work and slow company-wide coordination: Talgent’s stated initial vision is to give every employee a governed agent, then connect those agents so work is coordinated and the knowledge produced by finished work compounds inside the company. The repository describes Workbench as a self-hosted managed-agent platform and says it was built to keep the control plane, conversations, files and telemetry in the customer’s own cluster.
Target Markets
- Enterprises and organizations seeking governed AI agents
- Teams with cross-functional projects and coordination overhead
- Organizations requiring self-hosting, auditability and control of AI data and telemetry
- Developers and product teams embedding agent workflows through public APIs
- Individual employees using personal agents for real work
- Delegating research, writing, coding and document-production tasks to individual employees’ agents
- Running concurrent, resumable AI Works with files and repositories
- Automating agent-driven workflows from internal products and systems through Applications
- Coordinating cross-functional projects and routing changes to dependent teams
- Building organizational memory from completed project work
- Self-hosted enterprise AI where conversations, files, telemetry and model credentials remain in the customer environment