What is a BizMetry Telemetry Frame?¶

A BizMetry Telemetry Frame is a core abstraction within the BizMetry platform that enables the real-time capture of business telemetry data — representing one or more attributes of one or more resources (ASBOs).

Telemetry frames can be generated either automatically (via instrumentation and interception) or manually (via programmatic SDK calls).

Each frame acts as a data point set that encapsulates the state and context of a business event as it occurs.

🧩 Concept Overview¶

A Telemetry Frame represents a single observability event that BizMetry agents collect, process, and transmit to the central platform.

It contains all relevant data for that event, including which client generated it, what resources and attributes were affected, and what values were observed at that precise moment.

Each frame provides BizMetry with a fine-grained snapshot of business activity in motion, linking technical execution to business outcomes.

🧠 Structure of a Telemetry Frame¶

Every telemetry frame in BizMetry contains the following contextual components:

Component	Description
Timestamp	The precise moment in time when the frame was captured.
Client	The origin of the frame — representing the consumer or system invoking a resource.
Endpoint	The API or interface where the interception occurred.
Resources (ASBOs)	One or more business objects involved in the operation.
Attributes	Specific measurable characteristics of those resources (e.g., latency, cost, toxicity score).
Attribute Values	The actual numeric or non-numeric readings associated with each attribute.
Transactional Context	Metadata such as request ID, region, environment, correlation IDs, etc.

Self-Contained Representation

In BizMetry, every telemetry frame is a structured, self-contained representation of a business-relevant event or interaction.

🧱 Telemetry Frame Types in BizMetry¶

BizMetry supports the definition of multiple Telemetry Frame Types, each designed to model and capture a specific business scenario, operational context, or domain-level event.

A Frame Type acts as a logical template (or schema) that defines how telemetry should be collected for a given category of business interaction.
It specifies:

Which resources (ASBOs) are involved
Which attributes must or may be captured
What contextual metadata is relevant for interpretation and analysis

In essence, a Frame Type defines the semantic contract for a class of telemetry events.

🧠 Business-centric perspective (important clarification)¶

While a Frame Type is conceptually associated with a business action, this association is not necessarily 1:1.

A single Frame Type may be instantiated by multiple clients or channels, as long as they represent the same logical business intent and require the same telemetry semantics.

Frame Types model business meaning, not execution paths.

🔁 One Frame Type, multiple business entry points¶

For example, in a banking system, you may want to capture telemetry for a Funds Transfer operation with attributes such as:

Transfer amount
Currency
Risk score
Processing latency
Outcome (success / failure)

That same transfer can be initiated from multiple channels or platforms, including:

Mobile banking applications
Front-desk systems
Self-service kiosks
In-person assisted channels
External or partner platforms

Even though the originating action differs, the business intent and telemetry requirements remain the same.
All these entry points can emit the same Frame Type, enriched with contextual attributes such as channel, platform, or origin.

This enables BizMetry to:

Aggregate telemetry across heterogeneous channels
Compare performance, risk, and outcomes by channel
Preserve a unified, business-level view of the operation

🧩 Examples of Telemetry Frame Types¶

Frame Type	Purpose	Typical Metrics Captured
AI Chat Completion Frame	Observes AI model invocations and inference behavior.	Tokens in/out, latency, cost, model drift, bias indicators, toxicity score.
Order Transaction Frame	Tracks end-to-end execution of retail or e-commerce orders.	Order value, discount rate, payment latency, fulfillment time, failure rate.
Fleet Telemetry Frame	Collects operational metrics from logistics and delivery fleets.	Vehicle speed, fuel consumption, delivery delays, route deviation, geolocation.
Healthcare Visit Frame	Monitors patient appointment and care workflows.	Visit duration, waiting time, throughput, satisfaction index.
Finance Transaction Frame	Observes financial or banking operations.	Transaction value, response time, exchange rate, risk or fraud score.

🏗 Why Frame Types matter¶

By introducing Frame Types, BizMetry enables:

Consistent telemetry semantics across channels, platforms, and applications
Business-level observability instead of raw metric collection
Cross-channel aggregation and comparison without losing context
Extensibility, allowing new entry points without redefining telemetry models
Clear separation between what happened (business intent) and how it was triggered (execution path)

Frame Types are the foundation that allows BizMetry to translate technical signals into coherent, business-level insights.

Resource and Attribute Mapping

Each Frame Type is mapped to one or more Resources (ASBOs) and Attributes, ensuring metrics are always interpreted within the proper business context.

🧠 Frame Types within a Telemetry Template¶

Within a Telemetry Template, BizMetry supports the definition of multiple Frame Types, each dedicated to capturing different kinds of business interactions or operational scenarios.

Each Frame Type defines exactly which attributes to capture from the various resources that belong to the same template.

This modular approach allows organizations to model complex systems with multiple observability needs while maintaining clear separation of concerns between telemetry domains.

Multiple Frame Types Example

A single Telemetry Template for an AI project may define:

a ChatCompletionFrame for token usage, cost, and latency
a ModelEvaluationFrame for drift and bias metrics
a DatasetIngestionFrame for ingestion rates and validation errors

Each Frame Type encapsulates a specific subset of KPIs and resource–attribute mappings, keeping telemetry contextualized, efficient, and business-driven.

🧭 Telemetry Template Hierarchy Overview¶

Level	Description	Example
🧩 Telemetry Template	Root container defining observability rules for a Profile.	AI Observability Template
📦 Frame Types	Logical categories of frames capturing business events or KPIs.	`ChatCompletionFrame`, `ModelEvalFrame`
🧱 Resources (ASBOs)	Business objects being monitored within each frame type.	`AI Model`, `Prompt`, `Dataset`, `AI Agent`
📊 Attributes	Individual measurable fields captured from each resource.	`TokensUsed`, `Latency`, `CostUSD`, `SimilarityScore`

Refinement Hierarchy

Each level progressively narrows observability scope — from template, to frame types, to resources, and finally to attributes that generate actionable KPIs.

⚙️ Frame Generation Modes¶

BizMetry supports two complementary mechanisms for generating telemetry frames:

1. Automatic Instrumentation¶

BizMetry agents can automatically intercept relevant business operations — such as API calls, database queries, or AI inference requests — without modifying application code.

Frames are generated via auto-instrumentation hooks, minimizing effort and intrusion.

2. Manual Instrumentation (SDK-based)¶

Developers can manually emit telemetry frames using the BizMetry SDK when greater control or custom business events are required.

Schema Compatibility

Both methods produce frames compatible with BizMetry’s telemetry schema.

🚀 Transmission to the Platform¶

Telemetry frames are transmitted by BizMetry Agents to the central BizMetry Platform using the RMTP (Reliable Metric Transport Protocol).

Each agent:

Sends frames in batches to reduce overhead
Automatically retries on failures
Guarantees ordering and deduplication

Minimum Transmission Unit

The Telemetry Frame is the minimum unit of transmission in BizMetry.

🧮 Granularity and Selective Telemetry¶

Telemetry capture can be enabled or disabled dynamically at multiple levels:

Profile-level
Environment-level
Domain-level
Type-level (Client, Resource, Attribute)
Instance-level

This ensures performance efficiency, cost control, and contextual relevance.

⚙️ Agent Behavior and Performance Control¶

BizMetry Agents dynamically adapt batch size, frequency, and memory usage based on system load, ensuring telemetry never impacts application performance.

Scalable Observability

This distributed model enables high-fidelity, business-centric observability at scale.

📊 Example – AI Chat Completion Telemetry Frame¶

Resource	Captured Attributes
Prompt Template	Template ID, Language, Toxicity Score
AI Model	Name, Version, Hosting Type, Tokens In/Out, Latency
Datasets (RAG)	Dataset Count, Similarity Score
AI Agent	Cost (USD), Region, Client ID, Bias Deviation

Business Transaction Snapshot

This frame represents a complete business transaction snapshot for the AI interaction.

🖼️ Example — BizMetry Telemetry Frame Structure¶

🧭 Summary¶

A BizMetry Telemetry Frame is the atomic unit of business observability.

It captures real-time operational data at the intersection of Clients, Resources, and Attributes, enabling structured, actionable insights.

Key characteristics:

Automatic and manual capture
Multiple Frame Types per Template
Reliable RMTP transmission
Fine-grained configuration
Load-aware agent behavior