Buffaly Logo Buffaly
Examples

Execution timelines from
real Buffaly sessions.

These reconstructions show the runtime handling operational work, analyzing datasets, retaining reusable methods, and promoting repeated behavior into typed capabilities.

Autonomous Learning Arc

From Discovery to
Executable Code.

This timeline follows a task from contextual problem-solving to a retained reusable skill and then native ProtoScript code. It shows how ad hoc work becomes deterministic execution.

No Restart Required

The system internalizes logic and generates its own reusable capability in real time.

79.7% Lower Task-Specific Cost

Task-specific cost drops from 24,924 to 5,065, a 79.7% reduction in this measured session.

Open Learning Timeline
Runtime: FairPath_Learning_Session
OPTIMIZING...
Step 01. Discovery
Completes a novel task through the available tool surface.
Step 02. Reusable Skill
Method retained as a reusable skill that Buffaly can invoke again.
Step 03. Native Code
Method compiled to ProtoScript for deterministic execution.
[SemanticProgram.Deterministic]
function CalculateEligibility(Patient p) {
    // Now executing as native code in your VPC
    return p.Scores.Filter(s => s.Value >= 3).Count();
}
Global Orchestrator: HHS_Analysis_Log
FILE_SIZE: 11.2GB
# Analyzing 11GB Medicaid Provider Spending...
> Task: Segmenting 284k remote-care billing rows. [DONE]
> Task: Applying OIG statistical pattern risk scoring. [DONE]
> Task: Enriching 2,682 NPIs with Geocode context. [DONE]
Analysis Complete
TIME: 1 evening
Extreme Data Scale

11GB of Public Claims
Analyzed Overnight.

This session follows an HHS data export through a custom analysis pipeline that identifies anomalous provider behavior.

11.2GB
Source CSV
284k
Filtered Rows
Open Medicaid Timeline
HL7 import session

15k Patients for
Less than $13.

Buffaly took a real clinical archive, built and ran its own HL7 pipeline during the session, imported 15,075 patient records, and turned the result into a scored, payer-aware opportunity set for downstream action.

  • HL7 pipeline built during the session
  • 15,075 patient records imported
  • Scored, payer-aware output pipeline
Open HL7 Timeline

HL7 Workflow Snapshot

Imported Records
15,075
Imported Patient Records
Estimated Cost
$12.70*
Session Token Cost
HL7
ICD
Rx
* Estimated API-equivalent cost while building the import and qualification pipeline in a single session.

Continue into the runtime

Inspect the capabilities behind the timelines.

Browse skills Read the architecture