How do I enable a module?

Modules are enabled per-client via the dashboard or API. No code changes required.

Do modules share infrastructure?

Yes. Modules share the same tenant-scoped memory layer, so cross-module workflows (e.g. Deep Memory enriching retrieval that feeds into another module) work out of the box.

Can I use modules without the core TES API?

No. Modules extend TES — they require the core API and memory layer to function.

What's the difference between Deep Memory and Agent Memory?

They're the same thing. Deep Memory is the internal module name; Agent Memory is the product surface.

Are modules included in the free tier?

The free tier covers 1M proxied input tokens/month and basic memory features. Most advanced modules sit on Pro and Enterprise tiers — see the pricing page.

Hosted TES platform

Modules — opt-in features for the hosted platform

Name: Thing Event System (TES)
Author: Pentatonic

These modules ship as part of the hosted TES platform — analytics, memory, integrations that sit on top of the event spine. They are separate from the llm-proxy; the proxy works without them.

Just want to cut your token bill?

You don't need any of this. The llm-proxy is standalone — point your client at llm.api.pentatonic.com, add the X-TES-Token header, and you get memory injection on every request. Modules are for customers who want to extend the platform beyond what the proxy gives them out of the box.

Proxy quickstart Get a TES token

Open source

Connect via the AI Agent SDK

Modules are accessible via @pentatonic-ai/ai-agent-sdk or through the platform's HTTP API. Same tenant scoping, same memory layer.

Explore the SDK GitHub

7 modules

AI, analytics, and integrations

Deep Memory

deep-memory

Multi-layer persistent memory for AI agents. Episodic, semantic, procedural, and working memory with confidence decay and semantic search.

Memory NodesMemory Layers

Learn more →

Bias Pattern Evolution

bias-pattern-evolution

Detects recurring patterns and biases across agent traffic using Bayesian updates. Tracks pattern velocity, confidence, and lifecycle — emerging, stable, declining, dormant.

Bias Patterns

Learn more →

Predictive Modelling

predictive-modelling

ONNX-based models for demand forecasting, replenishment, inventory optimisation, and supply network analytics. Run predictions directly off ingested data.

PredictionsNetwork Maps

Pricing Engine

Rules

pricing

Rules-based and AI-powered pricing. Define pricing rules with conditions, or let the enrichment pipeline suggest market-based prices automatically.

Pricing ConfigPricing Rules

Conversation Analytics

Analytics

conversation-analytics

Track agent conversations, response quality, and interaction patterns across your customer-facing AI. Dashboard included.

Dashboards

Shopify Catalog Sync

Integration

shopify-catalog

Mirror your Shopify product catalog into TES with automatic AI enrichment. Products sync on a schedule, with embeddings generated for vector search.

Dynamic Yield

Integration

dynamic-yield

Sync your TES catalog to Dynamic Yield product feeds. Automatically push listed items to your personalisation engine with custom field mapping.

Frequently asked questions

Get started

Build on the TES platform

Install the SDK, enable the modules you need, and start emitting events.

Get API key View the GraphQL API