Thread vs Forethought
Side-by-side trajectory, velocity, and editorial themes.
Thread keeps deepening Magic's triage agent and tightening Voice AI controls for MSP shops.
Thread's recent cadence centers on iterative Magic AI releases (2.25 through 2.4) that broaden integrations (Hudu, Pia SmartForms), expand the triage agent into multi-step workflows, and polish the operator-facing emulator and settings UX. Voice AI is getting its own thread of refinements around overflow handling and per-agent contact mapping.
The product is being shaped into an integrated triage-and-resolution stack rather than an autocomplete-style assistant: the Magic agents are increasingly trusted to drive conversations to closure, route to the right knowledge source, and recover gracefully when a human is unavailable. Tooling for technicians (folders, resizable panels, emulator) is keeping pace so day-to-day operators can keep up with the agent's expanding scope.
Expect more knowledge-source integrations to land alongside the Hudu path, and the Pia SmartForms pattern to generalize into other PSA actions that the agent can run to completion. Voice AI will likely see additional overflow and handoff logic before a marketed 'agent handles the whole ticket' moment.
Forethought pivots from answering questions to executing outcomes via Orchestrator and Browser Agents.
Forethought is in the middle of a deliberate platform-narrative shift. April shipped two foundational pieces: Orchestrator, which routes business signals into deterministic AI actions across channels, and Browser Agents, which can take actions in apps that don't expose APIs. Test Suite landed alongside as the validation tooling for agent behavior before deployment. The CEO's 'Next Chapter' post frames the same direction in plain language: AI moving from answering to resolving.
The company is repositioning from a customer-support intent and triage AI to an outcomes-execution layer for enterprise customer experience. Browser Agents are the bet that the long tail of CX work lives in apps without proper APIs — making the agent capable of clicking through them is the moat. Orchestrator and Test Suite are the deterministic-control and validation pieces that make this defensible enough for enterprise procurement.
Expect a tightening of the integration story — pre-built Browser Agent flows for common CX systems like Zendesk and Salesforce Service Cloud — and an explicit outcomes-priced packaging emerging over the next quarter as the company moves past per-seat or per-resolution pricing.
See more alternatives to Thread →
See more alternatives to Forethought →