// 30-Day Pilot

Deploy in 20 minutes.
Understand your agent in 30 days.

Blocklog runs alongside your LangChain agent without touching production behavior — deployed hosted or within your own VPC. After 30 days, you have a forensic record of every AI financial decision your system made: what it decided, which inputs it used, how stale they were, and what would have changed the outcome.

// What you get

Four outputs. All from your own production data.

Every AI decision, preserved

A permanent system of record containing the evidence behind every AI decision your organization made during the pilot.

Every decision, explainable

Replay and investigate decisions using the exact context, actions, approvals, and governance controls that existed at the time.

Every decision, provable

Generate evidence packages suitable for auditors, regulators, customers, and internal investigations.

A decision infrastructure strategy

Know whether your organization can explain, verify, and govern AI decisions at production scale before broader deployment.

// Day by day

What happens in 30 days.

Connect your AI system

Deploy Blocklog alongside your agents — hosted by us or within your own VPC. No changes to production behavior. Decision capture begins immediately.

blocklog deploy --vpc

Capture real decisions

Every decision receives a permanent record containing inputs, retrievals, tool calls, approvals, and outcomes.

847 decisions captured

Replay critical events

Investigate failures, customer disputes, and unexpected outcomes using complete decision reconstruction.

replay(dec_9af1c18)

Generate evidence

Produce auditor-ready evidence packages directly from production decision records.

evidence_package.pdf

Evaluate the results

At the end of the pilot, you'll know exactly what your AI systems decided, why they decided it, and whether you can prove it later.

30-day pilot complete

// Good fit

The pilot works best when all four of these are true.

·

AI agent in production

Making autonomous financial decisions — refunds, fraud, credit, chargebacks, fee application, or payouts.

·

LangChain, OpenAI Agents SDK, or CrewAI

TypeScript SDK supports all three. Python SDK available for CrewAI and Celery-based stacks.

·

At least one engineer with 3–4 hours

For integration and feedback. The deployment itself takes under 20 minutes.

·

A compliance stakeholder who will read the report

Not required for deployment — but the pilot produces its most useful output when a compliance officer reviews the 30-day report.

// Not a fit right now

Honest disqualifications. Don't waste your 30 days on a wrong-fit pilot.

×

AI making recommendations only

No autonomous execution. If a human approves every action, the forensic record adds less value at this stage.

×

No financial decisions in scope

Blocklog is built for financial AI specifically. General-purpose agents are out of scope.

×

Test environment only

Blocklog produces meaningful output only from production traffic. Synthetic or test data produces a synthetic report.

// The deal

You run Blocklog for 30 days and give candid feedback. We give you a complete forensic report of your own agent's decisions — free, no obligation.

No sales follow-up during the 30 days. No pressure to convert. No pitch deck at the end.

At day 30, you get the forensic report regardless of what you decide next. The report is yours — generated from your production data, covering your actual decisions, formatted for your regulator.

The only thing we ask: give candid feedback on the replay engine and, if possible, have your compliance officer read the report and tell us whether it answers the questions their regulator is asking.

Three non-negotiables: AI in production making real financial decisions, an engineer with 3–4 hours, and at least one compliance-adjacent person who will read the report.

// Get started

Ready to see what your agent actually decided?

We review every application personally and reply within 48 hours. If you're a fit, you can be live the same day — hosted or in your own VPC.

Apply for the pilot

No sales follow-up during the 30 days. No obligation at the end. The forensic report is yours regardless of what you decide next.