Pisama
Pisama provides adaptive failure detection and automated healing for data workflows, matching known transaction-integrity patterns to a customer's own systems and terminology. The platform identifies likely failure locations from trace evidence, recommends fixes with supporting justification, and applies verified repairs directly to production workflows. It offers local Python and TypeScript SDKs, a self-hosted n8n server, and a hosted cloud preview with managed ingestion and browser-based visualization.
- Enterprise Software
- AI Agents
- Developer Tools
- Artificial Intelligence
- Software Only
Funding
Founders
Founder details are not available yet.
Product
Problem
Data workflows fail in ways that are specific to each organization's systems, integrations, and data vocabulary. Generic monitoring tools surface errors but lack the context needed to diagnose root causes across distributed transactions, leaving teams to manually trace failures and guess at fixes, which slows recovery and allows recurring issues to persist.
Solution
Pisama delivers adaptive failure detection that matches known transaction-integrity patterns—such as wrong-target charges and cross-system state drift—to a workflow's own systems and terminology. The platform identifies likely failure locations using trace evidence, recommends fixes with supporting justification, and applies them directly to production workflows after an automatic re-run verifies the resolution holds. Each verified fix becomes an update to the agent's instruction set, creating a closed loop where the system learns from its own failures and prevents recurrence. Pisama operates as a library-first product with Python and TypeScript SDKs, a self-hosted n8n server and dashboard, and a hosted cloud preview with managed ingestion and browser-based visualization.
Target Audience
Primary customers are data engineering and operations teams running complex workflow orchestration systems, particularly those using n8n, Python, or TypeScript, who need automated detection and healing of transaction-integrity failures without manual root-cause analysis.
Features
- Adaptive discovery maps a small set of known transaction-integrity patterns to the customer's specific workflow systems and vocabulary
- Evidence-based fix recommendations include the trace or system evidence justifying each proposed repair
- Verified healing applies fixes to live workflows only after an automatic re-run confirms resolution, with opt-in and reviewable changes
- Harness updates automatically extend the agent's instructions with verified fixes so the same failure does not recur
- Python SDK for local trace analysis, TypeScript local detectors and CLI, local MCP workflows, and framework adapters
- Self-hosted n8n product includes a purpose-built server, webhook and API-polling ingestion, SQLite storage, and a live execution stream
- Capability tiers distinguish heuristic detection, evidence-backed diagnosis, deterministic repairs, and model-generated fixes requiring a cloud key