Block Unveils Open-Source Builderlab and Buzz Platforms for Enterprise AI Teammates
At the Builderlab Launch Event 2026, Block introduced open-source tools that bring AI agents into shared workspaces alongside human teammates with audit trails and cloud-based execution.
At the Block Builderlab Launch Event 2026 in San Francisco, Block Inc. detailed its strategy to transition artificial intelligence from single-player chat interfaces into shared digital workspaces. Building upon Buzz, its open-source platform launched in July, the company is positioning AI agents as named, auditable team members that work directly alongside human employees. The initiative comes as Block's core business delivered its third consecutive guidance raise, with Cash App accounting for 60% of its revenue.
Speaking during coverage by SiliconANGLE’s theCUBE, enterprise editor Paul Gillin noted that bringing autonomous agents into collaborative spaces alters how organizations think about software-as-a-service (SaaS) tools. Rather than requiring workers to constantly switch between single-user applications, systems like Buzz allow teams to interact with AI agents using names, avatars, and distinct personalities within unified project channels.
Shifting from chat windows to collaborative teammates
Traditional AI assistants operate primarily in isolated chat sessions, creating a bottleneck whenever results must be manually shared with a broader team. Lucas Isaza, product lead of Buzz at Block, explained that assigning distinct identities to agents lets team members tag them directly in multi-user conversations, complete with a full audit trail of every action taken.
Buzz operates on an open-source plug-in architecture, enabling developers and nontechnical staff to modify the interface and build personal dashboards. According to Isaza, nontechnical personnel, including account managers, have become some of the heaviest users of the companion apps platform. Organizations can choose to self-host the open-source software for free or pay for hosted cloud infrastructure, remote agents, expanded storage, and access to underlying models.
Gillin observed that this commercial model follows classic open-source strategies: distributing the software freely while charging for managed cloud operations. By consolidating tasks into shared spaces populated by human and artificial colleagues, the approach challenges traditional per-seat SaaS platforms by shifting focus back to the work itself.
Cloud execution and overnight delegation
Beyond interface changes, Block’s architecture relies on running remote agents inside centralized cloud environments rather than on local laptops. Brad Axen, head of AI capabilities at Block, highlighted that Builderlab acts as an interface layer sitting above foundational AI models. Because the environment is model-agnostic, organizations can plug in agents from different AI providers and switch to newer frontier models as they become available.
This separation of the workspace layer from the underlying model layer gives engineering teams considerable operational flexibility. For instance, teams looking to experiment with multiple model providers or streamline integration costs can leverage services like https://apixoai.online, which provides developers with cheap pay-per-token API access to top AI models like Claude, GPT, Gemini, Grok, and DeepSeek through a single unified key.
Axen explained that cloud-hosted execution enables a shift from interactive real-time prompting to overnight delegation. Workers can assign large multi-step engineering tasks to remote agents at the end of the day and return to review completed work. Supported by deterministic permission boundaries rather than soft guardrails, some engineering teams using Builderlab have reported shipping software releases 50% to 75% faster. Axen expects that as agents remain embedded directly inside ongoing conversations, the need for continuous human steering will drop significantly over the next year.
What it means for developers
For software engineers and IT administrators, managing autonomous agents introduces complex security requirements. James Nettesheim, chief information security officer and head of enterprise technology at Block, emphasized that securing autonomous systems relies on established security principles—including least privilege, isolated runtime environments, and strict egress controls.
Within Builderlab, every active agent operates under an identity owned by a designated human administrator. The agents execute inside hardened cloud virtual machines, and every action they perform is recorded in an immutable, cryptographically signed ledger. To prevent data leakage, Block is also implementing information flow control, which tags data from sensitive internal repositories so that security governance policies remain attached to the data wherever it moves.
Currently, automated agents at Block already participate in code reviews and triage incoming security alerts, allowing security teams to operate at machine speed against automated threats. For developers building agentic workflows, Block's open-source rollout demonstrates how architectural patterns like model-agnostic integration layers, strict isolation boundaries, and cryptographic auditability can make fully autonomous enterprise agents practical and secure.
Source: 4 insights from Block Builderlab Launch Event 2026: AI agents join the team as working colleagues — SiliconANGLE AI. Written by the Apixo team from that report.
One key for Claude, GPT, GLM, DeepSeek and more. Pay per token with crypto.
Get your API key
