Synnada Inc.
Synnada is an AI infrastructure company building reliable, scalable, agent-native systems from prototype to production. Its current Recurse product is a serverless harness for creating, evaluating, and deploying custom specialist agents as tools, MCPs, or bots.
At a Glance
- Software and coding-agent teams
- Game studios and game-development teams
- Data-science and machine-learning teams
- AI infrastructure and platform engineering teams
- +1 more
AI Tools by Synnada Inc.
(1)
Recurse
Serverless Harness for Coding Agents
Discussions
No discussions yet
Be the first to start a discussion about Synnada Inc.
Latest News
Recurse published 'To solve or not to solve,' describing a game-solving agent and an iterative loop in which a simulator was rewritten between runs.
Recurse published 'No rulebook, no tells,' a case study on reverse-engineering 100 shipped levels and reaching 90% acceptance for generated levels.
Recurse published 'Make loops, not flows,' describing rebuilding a game-level-generation problem as an agent evaluated against player-facing quality.
Recurse publicly documented its current product, pricing, SDK/plugin repositories, deployment flow, and use-case catalog for custom specialist agents.
Products & Services
A serverless harness in which coding agents design specialist agents, tools, and verifiers; run alternatives; measure outcomes; refine candidates; and deploy the resulting agent as a tool, MCP, or bot. It supports one-time runs through the recurse CLI and ongoing MCP deployment.
Python package for authoring, packaging, and deploying Recurse applications.
Reusable agentic workloads with serverless execution that can be exposed as MCPs or run to completion.
A production-grade machine-learning compiler for efficient model execution.
Market Position
Synnada positions Recurse against general-purpose agent workflows that lack reliable evaluation and production execution. Its differentiation is an engineered loop around the model: domain tools, observable state, external measurement, explicit stopping conditions, stable contracts, and serverless deployment as MCPs/tools/bots. The broader Synnada platform is positioned for persistent, continuously operating agents rather than short-lived prototypes.
Leadership
Founders
Mehmet Ozan Kabak
Co-founder and CEO. LinkedIn describes him as an engineering professional with a broad Electrical Engineering and Computer Science background; Synnada identifies him as a creator of production AI/data infrastructure and a core contributor to Apache DataFusion.
Sami Can Tandoğdu
Co-founder and COO. He co-founded Synnada in December 2021; Synnada describes its founding team as engineers who have shipped production infrastructure and contributed to Apache DataFusion.
Executive Team
Mehmet Ozan Kabak
Co-founder & CEO
Engineering professional with a broad Electrical Engineering and Computer Science background; Synnada identifies him as a core Apache DataFusion contributor and creator of AI/data infrastructure.
Sami Can Tandoğdu
Co-founder & COO
Co-founder of Synnada and an engineer on the founding team that built production infrastructure and contributed to Apache DataFusion.
Founding Story
Synnada was founded in December 2021 by Mehmet Ozan Kabak and Sami Can Tandoğdu. Its initial vision was to build low-touch, self-contained machine-learning systems that produce robust, high-accuracy alerts and drive mission-critical actions in streaming-data environments, reducing alert fatigue; the company now frames that work as infrastructure for persistent, agentic systems.
Business Model
Revenue Model
Usage-based billing for active requests: model token usage plus secure serverless execution. Recurse says it adds no markup to model token rates.
Pricing Tiers
Usage-based model-call pricing.
Usage-based model-call pricing.
Usage-based model-call pricing.
Usage-based model-call pricing.
Usage-based model-call pricing.
Usage-based model-call pricing.
Non-preemptible direct runs cost 3x for function CPU and memory; model calls and other charges retain normal rates.
Target Markets
- Software and coding-agent teams
- Game studios and game-development teams
- Data-science and machine-learning teams
- AI infrastructure and platform engineering teams
- Organizations building continuously operating decision systems
- Procedural level design
- Game-play testing
- QA and chaos testing
- RNA sequence design and folding
- Query optimization
- Machine-learning model training and tuning