Start with the task and the permission boundary. A tool listing an action is not proof that it completes it reliably.
Documentation researched only. Pricing and eligibility vary by plan, region and workload; unresolved details are explicit. Performance has not been tested.
Anthropic describes Claude Code as an agentic tool used from the terminal for tasks such as code migrations and bug fixes.
Documentation researchedPricing: From $20/month (Pro, per Anthropic pricing page); usage limits apply
Best fit: Developers delegating repository changes from a terminal.
Comma describes a personal agent organised around ongoing Tasks, with connected apps, Mac access and a cloud computer.
Documentation researchedPricing: Price not verified
Best fit: Personal or team app tasks that continue in a cloud environment.
Cognition documents Devin as an autonomous software engineer that writes, runs and tests code, with an IDE, shell and browser.
Documentation researchedPricing: Free (limited); Pro $20/month; Max $200/month; Teams from $80/month minimum
Best fit: Engineering teams delegating well-scoped, verifiable tickets.
Intercom positions Fin as a customer agent across service and sales, combining skills and knowledge with customer history and context.
Documentation researchedPricing: From $0.99 per outcome; seat fees and minimums may also apply
Best fit: Customer-facing service and sales workflows.
Wajo describes Fo as an action agent for real-world bookings, coordination, purchases and life administration, with human help when AI cannot complete a task.
Documentation researchedPricing: Free plans advertised; real-world tasks use account credits
Best fit: Real-world coordination and life administration with vendor-assisted escalation.
Instinct describes a personal assistant reached by text or call, connected to applications and devices.
Documentation researchedPricing: Price not verified
Best fit: A messaging-first assistant connected to personal apps and devices.
Lucas describes a text-native iMessage assistant for email, calendar, documents, reminders and automations.
Documentation researchedPricing: Price not verified
Best fit: Personal administration through iMessage.
Manus describes agent tools with a browser and file system that can execute tasks and create deliverables, including presentations and websites.
Documentation researchedPricing: Credit-based; free daily credits documented; paid price unconfirmed
Best fit: Delivering task outputs rather than only answering questions.
Meta documents Muse as a personal agent for actions, reminders, ongoing conversations and file work.
Documentation researchedPricing: Limited free use documented by Meta; paid price not established
Best fit: Personal reminders, ongoing tasks and artifact work.
OpenAI documents dots using permitted plugins and shared ChatGPT memory, with data-access and action controls.
Documentation researchedPricing: Price not verified
Best fit: Personal connected-app work with shared ChatGPT context.
Poke documents an assistant in Apple Messages, Telegram, WhatsApp and RCS for email, scheduling, reminders and web search.
Documentation researchedPricing: Price not verified
Best fit: Messaging-based email, calendar and reminder work.
Replit documents an agent that creates applications, sets up projects, checks its work and fixes problems from natural-language instructions.
Documentation researchedPricing: Usage-based billing; no single price verified
Best fit: Building apps and related artifacts from natural-language requests.
Manus describes Cue as a standalone app for a personal team of agents on mobile and desktop, where each agent has its own email, phone number, wallet and computer.
Documentation researchedPricing: Price not verified; invitation code required
Best fit: People who want several agents with separate identities that hand work to each other, and who will review the final result.
Define success, the access you will grant, the approval points and how you will detect a failed task. A like-for-like decision needs those constraints before it needs a ranking.
Read the evidence method