With its commercial launch today, Swa Technology Inc. is betting on a new idea: harnessing the outputs of artificial intelligence agents from multiple AI model providers through one gateway that automates and orchestrates work.

The company said it designed a “foundational tool” that businesses can use to compare the outputs from leading models — including ChatGPT, Claude, Perplexity, Grok and more — to compare or add them into workflows. Users can also use outputs to validate and consolidate answers.

The company’s team and advisor network includes a dozen Nike Inc. alumni who helped lead digital transformation efforts at the global company. Swa’s mission is to provide a stable, adaptable platform. It aims to integrate the best tools and models for AI accuracy, resiliency and security.

“AI should feel like a utility, accessible to everyone, cost-effective, and secure,” said founder and Chief Executive Mike Sirchuk. “Swa gives businesses a simple way to roll out AI across the organization without forcing employees into a single model and without sacrificing the controls that security and IT teams need.”

From an operational viewpoint, the platform opens up support for cross-model workflows, meaning that employees can build AI agents that start with Anthropic PBC’s Claude to orchestrate task generation, then outsources its research and curation to OpenAI Group PBC’s ChatGPT for review, and finalize swith Codex for code generation or Google LLC’s Gemini for reports.

The company said it designed the platform to support the needs of all employees, from knowledge workers involved in adopting AI to veteran engineers. This includes security and information technology teams looking to reduce the use of shadow AI across their company, while controlling model access, data exposure and usage.

Finance teams also gain predictability through usage-based pricing to reduce per-seat waste and simplify vendor management.

The assistant can be accessed via numerous channels, including Slack, Microsoft Teams, WhatsApp and SMS — in public, private groups and direct messages. Swa says that means it can go everywhere employees can, providing them support and knowledge.

Employees can build agents within any interface, using @swa to summon the assistant and then tell it what kind of agent they need. After a few details, it will generate an AI agent fitting their needs. This could include a customer support assistant for policy-aligned responses to customer inquiries, a marketing copywriter, a research analyst or a conversational concierge. Each agent can be assigned to a particular model provider and given a name and a specific role.

Agents can also be put on “autopilot,” meaning they will run automatically at a certain time to execute a task without asking.

Image: SiliconANGLE/Microsoft Designer