Sevn AI Inc.
Sevn AI builds PySpur, an open-source platform intended to make AI agents reliable, transparent, and easier to build. PySpur lets AI engineers visually or programmatically build, test, debug, evaluate, and deploy agent workflows.
At a Glance
- AI engineers and ML engineers
- Startups and software teams building LLM applications
- Enterprise teams needing inspectable, testable agent workflows
- Organizations requiring self-hosted or on-premises AI infrastructure
- +1 more
AI Tools by Sevn AI Inc.
(1)
PySpur
Open Source AI Agent Builder
Discussions
No discussions yet
Be the first to start a discussion about Sevn AI Inc.
Latest News
PySpur repository merge updated the tiktoken dependency from 0.7.0 to 0.9.0.
Public promotion highlighted PySpur's no-code drag-and-drop agent builder, Python workflows, open-source model, package installation, repository, and documentation.
Age of LLMs profiled PySpur as an open-source graph-based LLM workflow editor, emphasizing visual building, node-level debugging, evaluation, JSON sharing, and self-hosting.
An independent technical post showcased PySpur's visual workflow editor, modular nodes, node-level debugging, workflow evaluation, Python/TypeScript stack, Apache 2.0 license, Docker deployment, and JSON workflow storage.
Products & Services
Apache-2.0-licensed visual playground and Python-based framework for building, testing, debugging, evaluating, sharing, and deploying AI-agent workflows. It can be installed with pip and run locally or on-premises.
Managed cloud deployment option for publishing agents and integrating them through APIs; the website also describes cloud and on-premises deployment paths.
Market Position
PySpur positions itself as an open-source, developer-oriented alternative to proprietary agent builders and workflow platforms. Its differentiators are a visual graph interface combined with Python extensibility, node-level debugging and evaluation, JSON/Git portability, self-hosting, and broad model/provider support. Public comparisons include ComfyUI, LangFlow, Flowise, Dify, Vertex AI Builder, Microsoft Copilot Studio Agent Builder, Lyzr Agent Studio, UiPath Agent Builder, and SmythOS.
Leadership
Founders
Jean Kaddour
AI researcher with a PhD in AI from UCL, more than seven years of ML-model-building experience, 1,400+ citations and 9 NeurIPS papers; previously worked at Reka and BCG Gamma.
Srijan Patel
ML engineer with more than ten years of experience building and deploying large-scale ML systems; previously an ML engineer at Meta and Head of AI at JioSaavn, which had more than 100M MAU; studied Electrical Engineering at IIT Kanpur.
Executive Team
Jean Kaddour
Co-Founder & CEO
AI researcher; PhD in AI at UCL; prior experience at Reka and BCG Gamma; author of nine NeurIPS papers.
Srijan Patel
Co-Founder & CTO
ML engineer; former Meta ML engineer and JioSaavn Head of AI; Electrical Engineering graduate of IIT Kanpur.
Founding Story
The founders launched a graphic-design agent in early 2024 that quickly reached thousands of users. They found existing agent reliability and debugging tools inadequate, so they built PySpur as the playground they wished they had for testing workflows, inspecting failures, and iterating toward production reliability.
Business Model
Revenue Model
Open-source Apache-2.0 software is available for self-hosting; the company also offers a managed cloud service with tiered plans and an enterprise sales model.
Pricing Tiers
For tinkerers and hobbyists; 1 team member, 10 apps, 1GB vector storage, 50 documents, and 500 message requests per day.
Unlimited team members and apps, 100GB vector storage, 10,000 documents, and unlimited message requests.
SSO via SAML/OIDC, multiple workspaces, deployment assistance and SLAs, model load-balancing, and dedicated Slack/email support.
Target Markets
- AI engineers and ML engineers
- Startups and software teams building LLM applications
- Enterprise teams needing inspectable, testable agent workflows
- Organizations requiring self-hosted or on-premises AI infrastructure
- Research and education users working with multimodal or RAG systems
- Building and iterating on production AI agents
- Prompt and workflow experimentation
- Debugging multi-step agent failures
- Evaluating agents against real-world datasets
- Document and knowledge-base RAG applications
- Multimodal agent workflows