Seattle startup Read AI has launched a new AI-powered assistant called Ada, branded as a “Digital Twin,” that works through email to help users schedule meetings, answer questions, and draft replies. The company says Ada is available to all users starting Thursday and can be configured by emailing “ada@read.ai” with the subject line “Get me started.” Read AI plans to expand Ada to Slack and Microsoft Teams in the near future.
Ada builds on Read AI’s existing meeting notetaker and productivity tools, positioning the assistant as more than a scheduling bot. The company describes it as a digital twin that can act on a user’s behalf around the clock. Users can cc ada@read.ai on an email thread and ask it to find time on calendars, respond to scheduling conflicts, draft replies, or pull information from connected systems.
When asked to coordinate a meeting, Ada replies in-thread with the user’s availability. If the recipient suggests different times, Ada continues the negotiation automatically. While it has access to a user’s calendar through Read AI, it does not disclose details about other meetings. For more sensitive actions, the assistant “sidebars” with the user first — preparing draft responses and waiting for approval before sending anything externally.
Beyond scheduling, Ada can answer questions using a company’s knowledge base, prior meeting transcripts, connected files, CRM data, and public web searches. For example, users can ask for updates on quarterly goals, and Ada will compile relevant information from past discussions and connected systems. If someone else asks a question in an email thread, Ada can prepare a response for the user to refine before it goes out. The company says Ada will not share sensitive information without explicit permission.
Read AI says its platform integrates with more than 20 native services and draws, on average, from about 10,000 documents per user. Unlike some AI tools that rely on model context protocols (MCPs) to connect to external services, Read builds a knowledge graph from meeting data and connected apps to provide contextual responses.
CEO David Shim framed the launch as a shift from “AI assistant” to something closer to a software colleague. In internal beta testing, he said roughly a quarter of user interactions with Ada were simply to say “thank you,” which he views as evidence that users treat it more like a teammate than a tool. He described the product as moving Read AI from “a system of record for productivity” to an “extension of you.”
Shim also said the release marks a transition from AI that waits for instructions to AI that can act more proactively. Over time, Ada is expected to take additional steps on its own — for example, prompting users to follow up on action items mentioned in meetings.
Founded in 2021, Read AI has raised more than $80 million in funding and now reports over 5 million monthly active users. The company says it sees about 50,000 new sign-ups per day and serves a broader audience of around 100,000 users who consume meeting summaries without creating an account. While 60% of its users are outside the U.S., revenue is roughly split evenly between domestic and international markets.
Read AI enters a crowded field of AI agents and workplace copilots from companies like Microsoft and Google, along with startups focused on inbox management and autonomous task execution. Other meeting-focused tools are also expanding beyond note-taking. Granola has introduced repeatable prompts to surface knowledge from meeting data, and Quill recently launched with integrations into tools like Linear, Notion, and CRMs to automate tasks.
By centering Ada inside email and tightly linking it to meeting and document context, Read AI is betting that the inbox remains the control center of workplace communication. Whether users are comfortable letting a digital twin handle increasingly larger portions of that flow will determine how far these agent-style assistants can go.
This analysis is based on reporting from GeekWire.
Image courtesy of Read AI.
This article was generated with AI assistance and reviewed for accuracy and quality.
