Your AI agent is ready for enterprise customers. Can you prove it’s secure?
Rends implements and tests agreed agent controls and prepares the evidence your auditor will ask for — verifiable offline, without access to Rends systems — in 3 weeks.
From $15,000 · 3 weeks · $2,500 deposit, fully refundable until Week 1
Fed leads every sprint personally, so we start at most 2 sprints a month.
See the proof before you buy
View a sample evidence bundle · Verify a demo chain yourself
Sample uses synthetic data and a sample key.
Who it’s for
Companies whose AI agents take real actions for customers (support, coding, finance, operations) and who are preparing for AIUC-1, SOC 2 or an enterprise security review.
What you get
- Week 1 — Control-gap matrix. We map your agent against AIUC-1’s agent-action, tool-call, oversight and logging requirements. You receive a control-gap matrix: each requirement, the evidence you have today, and what’s missing.
- Week 2 — Agreed controls, implemented and tested. Together we agree which controls you need, then implement and test them on the Rends platform: policy checks before your agent acts, approval holds for high-risk actions with each approval recorded, tamper-evident decision logs, and an emergency stop.
- Week 3 — Evidence bundle, test results and auditor handoff. An evidence bundle of your agent’s decisions, mapped to the controls and verifiable offline without access to Rends systems, plus the Week 2 test results and a handoff to your auditor.
What your engineering team gets back
- A control-gap matrix for one agent workflow
- Agreed controls, implemented and tested
- Test results showing the authorization boundaries
- An exportable evidence bundle with offline verification instructions
- A technical handoff your auditor can review
See a sample evidence bundle → Sample, synthetic data, signed with a sample key, not a customer bundle. Sample public key
Why teams choose a sprint
- Your enterprise customer or auditor has a date. Building agent controls and evidence in-house competes with your roadmap.
- One bounded engagement: controls implemented and tested on Rends, plus the evidence your auditor will ask for.
- A fixed scope, a price you can see, and a written promise to fix our work.
What your auditor receives
- A signed audit bundle: your agent’s decision records in hash-chained order, signed by Rends.
- Rends’ published verifier script and public signing key, so the bundle can be checked offline without access to Rends systems.
- A control map showing which records evidence which control.
- The Week 2 test results.
Why Rends
- Live runtime agent controls. Policy checks before an action runs, approval holds with each approval recorded, and an emergency stop are running in production today.
- Published open specs. RPL, the policy language Rends evaluates, and AALS, an open record format for agent decisions, are published in public GitHub repositories.
- Evidence checked offline. Your auditor checks the signed bundle with our published verifier, offline, without access to Rends systems.
What we need from you
- A technical owner for the agent, available a few hours a week during the sprint.
- A walkthrough of what your agent does: the actions and tools it can use, and who approves what today.
- A staging environment where we connect your agent to Rends through our Python or TypeScript SDK, or our API.
- The security evidence you already have, such as a SOC 2 report or security policies.
- Your auditor’s contact, if you have one, for the Week 3 handoff.
Week by week
- Before week 1: Reserve with the refundable deposit, or book a readiness call first. Either way, we confirm scope at kickoff.
- Week 1: Kickoff, a walkthrough of your agent, and the requirement mapping. Ends with the control-gap matrix.
- Week 2: We agree the controls, connect your agent to Rends in staging, implement them and test each one. Ends with the test results.
- Week 3: We export the evidence bundle, check it offline with you, map it to the controls and hand it to your auditor. Ends with the handoff session.
3-minute AIUC-1 readiness check
13 quick questions about how your AI agent is controlled today. You’ll see which areas look covered and where there’s likely a gap. It’s based on your answers, not an audit.
Based on your answers, not an audit. Not affiliated with AIUC.
A. Actions checked before they run (AIUC-1 B006)
B. Limits on tool calls (AIUC-1 D003)
C. Approval for high-risk actions (AIUC-1 D003.4)
D. Stopping the agent (AIUC-1 C009, optional)
E. Ownership and accountability (AIUC-1 E004, E015)
F. Logs an auditor can check (AIUC-1 E015, E015.4)
Answer all 13 questions to see your results.
Price
From $15,000. The $2,500 deposit is credited to the total.
From $15,000 for one agent workflow and one supported integration. Additional workflows and custom integrations are quoted separately.
Fed leads every sprint personally, so we start at most 2 sprints a month.
Deposit is credited to the total and refundable in full until Week 1 begins. Full price confirmed in writing at kickoff. The remaining balance is invoiced after the kickoff call confirms scope, and is due before Week 1 begins. If your setup doesn’t fit the sprint, the deposit is refunded in full.
Led by Fed Alcius, founder of Rends.
Our promise
If an agreed deliverable doesn’t meet its written acceptance criteria because of our work, we fix it at no extra fee. We don’t promise certification or an auditor’s opinion.
Questions
- Is the deposit refundable?
- Yes, in full if you cancel before Week 1 begins. Once work starts, it’s credited to the work done.
- Which agent setups do you support?
- Agents that can ask Rends for a decision before they act: through our Python SDK (rends-agent-sdk, with adapters for LangChain, LangGraph, CrewAI, AutoGen, OpenAI Agents, Google ADK, Haystack, LlamaIndex and PydanticAI), our TypeScript SDK (@rends/agent-sdk, with LangChain, OpenAI Agents and Google ADK adapters), or our REST API from any language.
- Does the sprint include the AIUC-1 audit?
- Certification is your auditor’s decision, not ours. We prepare the agent controls and the evidence for the requirements listed above; we do not cover every AIUC-1 requirement.
- How is our data handled?
- Rends receives what your agent sends with each decision request and records each decision in a hash-chained, signed audit trail, under the terms of our privacy policy. The evidence bundle is exported to you.
Your auditor decides certification; we prepare you for it. We cover the agent-control requirements listed above, not every AIUC-1 requirement.