Merlinby RLM-Labs

Confidence is cheap.

Every answer comes with the receipts.

Merlin gathers evidence, argues with its own conclusions, and refuses to claim work it can't prove. It runs on your hardware.

Read the rulebook.
From the operating record

24 of 25 tasks complete.

Full-success answer rejected

task_total
25
task_completed
24
task_in_progress
1
committed_writes
11

One task was still in progress. The system refused to call the run complete.

Open the record

Historical production excerpt. Identifiers omitted.

01 / How it works

The work has a structure.

Plan. Gather evidence. Challenge the findings. Verify the result.

A request enters Merlin's plan, analyze, execute and verify workflow. Evidence and context, model routing, and tools and workspace connect to the orchestrator. Model routing connects to a GPU inference fleet and optional cloud fallback.
Conceptual architecture · not every request follows every stage. GPU symbols are representative.View full size ↗

02 / An evidence-governed AI runtime

The model reasons. The runtime holds it to account.

Merlin brings models, tools, evidence and verification into one system. Its defining responsibility is deciding what that system has earned the right to claim.

01 / Evidence

Start with something real.

A file read. A command result. A source retrieved. The evidence record gives a claim something to stand on.

Read the requirement

02 / Challenge

Make the finding answer back.

Findings move through rebuttal and verification. Refuted and inconclusive are first-class outcomes.

Read the requirement

03 / Authority

Completion has to be earned.

A proposed success answer can be rejected when tasks are still open. The model does not have the final word.

Read the requirement

04 / Ownership

Keep control of the boundary.

Run on infrastructure you control. Choose the models, the routing policy and whether cloud escalation is allowed.

See the boundary

Rules, in the system’s own words

Binaries are never evidence.Capacity is not spend.Prior claims are context, not proof.

03 / Your infrastructure

It runs where you put it.

Local execution keeps change and answer authority. Cloud escalation is configuration-gated and can be switched off. The boundary should be a decision you can inspect.

Explore the local boundary ↗