Inside Cursor, Colossus, and Grok Bot With Nick Miller

Guest
Nick Miller
Lead Field Engineer
@
SpaceXAI

Nick Miller, Lead Field Engineer at SpaceXAI, joins us in the studio at Agent Conf 2026 in Warsaw to discuss Cursor's move into SpaceXAI and the engineering experience behind Grok Bot.

We talk about how access to Colossus changed for Cursor's model team, how the companies' products fit together, and the browser capabilities that enable agents to act on websites. Nick also explains how the team drew on its work at Cursor to take Grok Bot to an internal beta in about a month.

Cursor's place within SpaceXAI

Nick describes a relationship that began with access to Colossus for model training and grew into an acquisition. Cursor brought its coding tools and desktop agent experience; SpaceXAI brought its broader AI products and infrastructure.

At the time of the interview, the teams were still working through questions about their brands, subscriptions, and a unified platform. Nick explains why their different product focuses gave them room to collaborate.

What Colossus changed for model training

Access to compute had limited the model experiments that Cursor could run. Nick explains how Colossus gave the team more capacity to train models and explore different approaches.

He describes the progression from Cursor's Tab autocomplete model and code-focused Composer models to collaboration on general-purpose Grok models. We also discuss whether adding compute and training data will keep improving model capabilities, a question Nick says remains open.

The Cursor experience behind Grok Bot

The team had already spent years building Cursor's agent harnesses, the software that connects models to tools and manages how they act. Its experience with computer use included a built-in browser in Cursor's desktop app.

Nick says that accumulated experience contributed to Grok Bot reaching an internal beta in about a month. He explains how earlier engineering work carried over to the new product's browser capabilities.

Browser interactions alongside APIs and MCP

Grok Bot's bots run in an isolated virtual machine environment. Nick describes how that controlled environment fits into the team's approach to computer use.

The team wanted bots to handle work through whichever interface was available: an API, a Model Context Protocol (MCP) server, or the browser interface a person would use. We ask what went into that browser experience and why it was a priority during development.

About the guest

Nick Miller is Lead Field Engineer at SpaceXAI. At Agent Conf 2026, he presented "Building a Software Factory with Cursor, xAI, and Grok Bot." This studio interview follows up on the company relationship and the engineering behind the tools.

Resources

Insights

Learn more about AI

cover

Nov 17Talk

Efficient On-Device LLMs in React Native Done Right

Artur Morys-Magiera will cover efficient on-device LLMs in React Native, including runtimes, model selection, quantization, and platform constraints.

cover

Nov 17Talk

On-Device AI After WWDC: What Changed for React Native

Mike Grabowski will explain what changed for system-provided on-device AI after WWDC and what React Native developers can build with it.

cover

Nov 16Talk

Beyond Coding Agents: Making ADLC Work Across Platforms

Mike Grabowski will explain how to close the full agentic development lifecycle across mobile, web, and native platforms.

Building the React Native Software Factory — cover

Nov 11Talk

Building the React Native Software Factory

Join us at React Foundation Meetup London on November 11 to explore how AI agents build and debug React Native apps, from one terminal to the cloud.

Building a Software Factory: Write Specs an Agent Can Build From — cover

Nov 6Article

Building a Software Factory: Write Specs an Agent Can Build From

Learn how to research requirements, resolve open decisions, and write testable specs that AI coding agents can implement without the planning conversation.

 Large-Scale Migrations With Coding Agents: Planning, Verification, and Rollback — cover

Nov 4Article

Large-Scale Migrations With Coding Agents: Planning, Verification, and Rollback

Plan codebase migrations with coding agents using reviewable requirements, verification checks, staged rollouts, and defined rollback criteria.

Quantization Impact on LLM Quality: A Case Study of Apex — cover

Nov 4Talk

Quantization Impact on LLM Quality: A Case Study of Apex

Artur Morys-Magiera and Lech Kalinowski share findings from 344 benchmark runs with Apex, our Gemma-based LLM, examining the trade-offs between model size, decode speed, and quality.

Conversational Banking: From Customer Questions to Authorized Actions — cover

Nov 2Article

Conversational Banking: From Customer Questions to Authorized Actions

Explore conversational banking architecture through a card-dispute example, covering tool access, authorization, execution, observability, and human handoff.

Building a Software Factory: The Night My Agents Waited for Me — cover

Oct 28Article

Building a Software Factory: The Night My Agents Waited for Me

Learn how permissions, review limits, and missing product decisions stalled two overnight coding agent runs, and how their maintainer revised the workflow.