Reviews, comparisons and guides for AI tools for freelancers and solopreneurs.
Short answer: IrisGo keeps your screenshots, transcripts, and recorded workflows on your own computer, but any task you send to a cloud model travels through OpenRouter to OpenAI, Anthropic, or Google, and its firm no-training promise covers only your Google data. Our verdict is USE WITH CAUTION for client work. Policy reviewed October 2026, a…
Short answer: Willow Voice does not keep your audio or dictated text by default, but every dictation is processed in the cloud, and the new Knowledge feature asks you to hand over personal context that the privacy policy does not yet describe. Our verdict is USE WITH CAUTION for client work. Policy reviewed October 2026,…
Short answer: SpeechShield keeps your transcripts and client files encrypted on your Mac, but every live session sends audio and meeting context to an AI provider, and the app is built to stay invisible to the people you are talking to. Our verdict is USE WITH CAUTION for client meetings: run it in local mode…
Short answer: Cue by Manus gives each AI agent its own email address, phone number, wallet, and cloud computer, and lets incoming emails or Slack messages trigger those agents with no human check in between. Our verdict is AVOID FOR CLIENT WORK until Manus publishes Cue-specific data controls and approval settings. Policy reviewed October 2026,…
Short answer: OpenAI’s Dots can read your connected inbox, calendar, and Slack around the clock, and you cannot delete a single memory it builds from that client data without wiping the whole agent. Our verdict is AVOID FOR PAID CLIENT WORK on personal ChatGPT plans until OpenAI ships per-item memory controls. Policy reviewed October 2026,…
Short answer: Earlyn keeps your screen history and meeting transcripts encrypted on your own Mac and sends nothing to its developer, but it still captures every client document you open and every call you record. Our verdict is USE WITH CAUTION: a strong local-first design that becomes risky the moment you connect a cloud model…
Short answer: Jotform Sign for ChatGPT and Claude moves your client contracts through two companies at once, and Jotform’s own AI policy says content you put into form fields may be used to train its AI. Our verdict is USE WITH CAUTION: fine for client agreements once you lock down training settings, risky if you…
Short answer: Arc sends whatever is on your screen, plus any call recordings you feed it, to Google’s Gemini API, and its policy never says whether that traffic runs on Google’s paid tier or the free tier Google may train on. Our verdict is AVOID FOR PAID CLIENT WORK until Arc discloses that. Policy reviewed…
Short answer: on Free, Pro, and Business plans, Superhuman Go can use your writing to train its AI models unless you switch that off, and it reads the full content of the emails, docs, and connected apps it works in. Our verdict is USE WITH CAUTION: fine for client work once training is off and…
Short answer: on the free tier, Google AI Studio uses your prompts, uploaded files, and outputs to improve Google’s products, and human reviewers may read them, so client material does not belong there. Our verdict is USE WITH CAUTION: acceptable for client work only inside a billing-linked project, never on the free tier. Policy reviewed…
