Turning usage data into revenue forecasting

With accurate maps and weather information, captains are more likely to steer their ships through choppy waters safely. Similarly, with more accurate revenue forecasts, business leaders are more likely to guide and grow their businesses safely through challenging conditions.

With the rise of consumption-based pricing models, a renewed focus on the usage data is needed to have predictable models. In this session we will demonstrate how to forecast revenue for usage-based applications, using a modern datalake approach. We will compare multiple ML models to forecast consumption for various common use cases such as compute hours, storage, API calls, and monthly active users (MAUs).

UCP 101 (but actually useful): “Commerce Primitives for Agents”

The Universal Commerce Protocol (UCP) is an open standard that defines a common language and functional primitives so AI agents, merchant systems, payment service providers, and credential providers can interoperate across the full shopping journey—discovery → checkout → post-purchase—without bespoke, one-off integrations.
This session breaks down UCP’s core concepts and “commerce primitives,” then shows how teams can adopt UCP incrementally on top of existing retail infrastructure and pair it with agentic payment flows (e.g., AP2) when payment authorization and safety are required.

Building anInteroperable Agentic Stack: Lessons from MCP, NLWeb, and UCP

As agents move from “chat” to real-world action, we need an interoperable stack that spans discovery, capability execution, and transactional workflows—without bespoke integrations per site or platform. This talk distills practical lessons from three emerging building blocks: NLWeb (a way for websites to expose natural-language interfaces using existing web semantics like Schema.org/RSS and an MCP-compatible server surface), MCP (a standardized way for agents to invoke tools and retrieve context), and Google’s Universal Commerce Protocol (UCP) (an open-source set of common commerce primitives meant to standardize end-to-end shopping journeys across merchants, surfaces, and payment providers).

From Scripted to Smart: How to Systematically Humanize Your AI Sales Agent

Over the past few years, we've focused on one goal for our Sales AI agent: making interactions feel more human.
But what does “human” actually mean? For some, it’s about natural language. For others, it’s about empathy or tone.

In this session, we’ll share how we integrated human reviewer feedback into our training loop—what worked, what didn’t, and how we measured success.
We’ll also dive into evaluation: how do you actually test if an AI response feels human?

From Scripted to Smart: How to Systematically Humanize Your AI Sales Agent

Over the past few years, we've focused on one goal for our Sales AI agent: making interactions feel more human.
But what does “human” actually mean? For some, it’s about natural language. For others, it’s about empathy or tone.

In this session, we’ll share how we integrated human reviewer feedback into our training loop—what worked, what didn’t, and how we measured success.
We’ll also dive into evaluation: how do you actually test if an AI response feels human?

Billing infrastructure for developers- the hard way

Building a data pipeline for usage-based billing seems easy at a first glance. A developer can write a cron job to query the database at the end of the month and send the total amount to the payment provider. Although it can begin simply like this, the complexity rapidly grows under the weight of real-world billing requirements.

From Tool Calls to Revenue: Metering and Billing MCP Servers at Scale

Building an MCP server is getting easier. Monetizing one is not.

At first, usage-based billing looks simple: count tool calls, aggregate them monthly, and send an invoice. But once MCP servers move into production agent workflows, the definition of “usage” becomes much harder. Was that one tool call or ten? Should you bill for tokens, API calls, records retrieved, compute time, successful outcomes, or prepaid credits consumed? What happens when an agent retries, chains tools, streams results, or calls the same resource through multiple clients?

In this session, we’ll dive into the infrastructure behind trustworthy metering and pricing for MCP-native products. We’ll cover how to design a billing pipeline that can:

Count usage once and only once, even with retries, failures, and agent loops
Handle high-volume event streams from MCP tool calls
Give customers real-time visibility into current-period usage
Support pay-as-you-go, prepaid credits, drawdown models, and hybrid pricing
Preserve developer trust by making usage explainable and auditable

MCP is often described as a standard way for AI clients to call tools, resources, and prompts across servers. That makes every MCP server a potential commercial surface area: tool calls become billable events, resources become metered access, and agent workflows become revenue streams