EmbarkNowAn AI workspace for mission-driven organizations—bringing everyday operations, specialist companions, and human-reviewed decisions into one connected product.
Explore the live productFor a mission-driven organization, a donor conversation, a volunteer need, and a board decision are parts of the same story. But the work often lives in separate inboxes, spreadsheets, documents, and systems. Recognizing what matters—and deciding what to do next—takes time away from serving people.
EmbarkNow brings those activities into a shared workspace. The challenge was to make a broad AI platform understandable in daily use: surface the right context, prepare useful next steps, and keep people in control of consequential decisions.
I built the project end to end, connecting the application experience to the APIs, data layer, and agent workflows behind it. My scope covered the public product experience, authenticated workspaces, operational modules, AI companion interfaces, and the systems that route work and track decisions.
The command center gives the organization a daily starting point. Activity from modules is organized into attention items, work in progress, and next actions. A companion sits beside that operating picture, so guidance has a clear relationship to the work on screen.
The module catalog organizes capabilities across relationships, finance, communications, governance, and AI. Access labels and work-unit filters help users understand which tools they can explore, which are included, and which consume usage. A common navigation system connects the modules without erasing their different jobs.
In Donor Relations, the work is grounded in a weekly plan: task progress, thank-you follow-ups, committed activities, and relationship priorities. The companion remains available beside the workspace. This connects conversation to the specific plan or record the user is working with, instead of making them reconstruct the context elsewhere.
Financial stability, discernment, community development, and fundraising have different contexts. Specialized companions bring those responsibilities into separate roles. The interface shows the selected companion’s module access and estimated work-unit cost, with expanded access represented as a request that needs approval.
A recommendation and an approved action are different states. The Approval Inbox makes that distinction visible across submitted planning revisions and draft communications. Pending, approved, declined, and deferred work stays organized, while reviewers can inspect the material before deciding what should happen next.
The application, conversation layer, agent workers, and operational data work together behind the interface.
Modular screens and contextual companion panels, with shared navigation, reusable interface components, and client-side data fetching.
Typed request handling and conversation endpoints connect the web application to operational and AI workflows.
Routed requests reach a worker that resolves the relevant module crew, executes its work, and publishes progress and results.
A shared database layer supports application context and execution records, with observability contracts linking work to its tenant and lifecycle.
A pack-based crew factory loads a module’s implementation behind a common execution interface. Operational areas share infrastructure while keeping their own tools, context, and responsibilities.
Redis-backed workers separate agent execution from the web interface. Progress and result events give the application a way to represent work that continues beyond a single request.
Approval states, reviewer context, and access maps are part of the interface as well as the backend model. Users can distinguish suggested work from reviewed decisions and understand the scope of a companion’s guidance.
The result is a live, connected product that spans discovery, daily operations, specialized AI assistance, and human review. The platform brings a 19-module catalog and four specialized companions into a common application experience, with a detailed donor workflow that connects planning to follow-through.
Useful AI depends on the workflow around it: the context it receives, the work it produces, and the point where a person makes the decision. EmbarkNow brings those pieces together in the same product.
Screenshots captured from the live application. Account details and individual donor records are excluded from the selected views.