SpendNod is free and open-source — self-host it yourself, or let us run it for you. We offer managed deployments, white-glove onboarding, and enterprise SLAs so your team can focus on building agents, not infrastructure.
Every agent making financial decisions without human oversight is an unmitigated risk to your organization.
LangChain, CrewAI, AutoGen - none provide built-in authorization logging. When your CFO asks "who approved this transaction?", most teams have no answer.
One AI agent overspending on a client account, one unauthorized purchase, one unexplained charge - that's a client lost and a lawsuit started. The cost of oversight is zero compared to the cost of a mistake.
Percentage of global annual revenue for failing to demonstrate human oversight. Even US companies with EU customers or operations are in scope. Enforcement begins August 2, 2026.
Every authorization request, every rule evaluation, every human decision - timestamped, attributed, and exportable. Complete audit trail for compliance, insurance, and client trust.
Whether you're preparing for the EU AI Act, satisfying SOX auditors, or protecting client trust - SpendNod provides the governance layer.
Every transaction has an auditable decision chain. When a client asks "why did your AI spend $5,000?", you have the proof - rules, evaluation, and who approved it.
Every authorization logged with timestamp, user identity, parameters, rule evaluation, and outcome. Exportable for auditors, insurers, or legal review.
Configurable rules engine with per-agent, per-category, per-vendor controls. Enforce organization-wide spending policies across all AI agents.
Transactions above configured thresholds require explicit human approval. Approvals attributed to specific users with full context - the oversight that fiduciary duty requires.
Auditable proof that AI systems operated within authorized parameters with human oversight. Required for organizations with EU customers or operations by August 2, 2026.
Works with any AI agent framework via MCP or REST API. One governance layer for LangChain, CrewAI, AutoGen, and custom agents.
SpendNod integrates with your existing agent infrastructure without code changes.
Add the SpendNod MCP URL to your agent configs or integrate via REST API. Each agent gets its own API key and policy scope.
Set organization-wide spending thresholds, approved vendor lists, category restrictions, and daily velocity limits. Start with templates or build custom rules.
Every agent transaction is evaluated against your policies in real-time. Flagged transactions go to human reviewers. Everything is logged.
SpendNod is free and open-source. If your organization needs managed hosting, compliance reporting, or enterprise SLAs — we can help.
No. SpendNod is an authorization layer. We never process, hold, or transfer funds. We provide the decision (approved, pending, denied) and the audit trail. Your existing payment infrastructure handles the actual transaction.
Three reasons. First, liability - if your AI agent overspends on a client account, you need proof that it was operating within authorized parameters. SpendNod provides that proof. Second, if you have any EU customers or operations, the EU AI Act applies to you (enforcement August 2, 2026). Third, insurers and auditors are starting to ask about AI governance controls. Having SpendNod is a checkbox on applications and audit reviews.
For MCP-compatible agents: add one URL to your config file. For REST API integration: one API call before each transaction. Most teams are live within a day, not weeks.
If SpendNod is unreachable, your agent receives a timeout error. The agent should be configured to fail closed - meaning it does not proceed with the transaction. This is the safe default and is required for compliance.
Yes — that's the default. SpendNod is open-source. Clone the repo from GitHub at github.com/gstack-ds/spendnod and deploy it on your own infrastructure. If your team would prefer managed hosting, contact us and we'll handle the deployment.
SpendNod works with any agent that can make an HTTP call or connect via MCP. This includes Claude, ChatGPT, LangChain, CrewAI, AutoGen, Microsoft Agent Framework, and custom agents. If it can call a URL, it works with SpendNod.
Self-host from GitHub, or contact us for managed hosting, enterprise support, and compliance assistance.