hi! we're The Orchestration Company (Dcouple, Inc).
we build ai-native tools for the new way work gets done: orchestrating — people describing outcomes, agents writing and changing software, and teams needing trusted systems around that work.
pane is for the software you ship.
doozy is for the work you need done.
morph is for the internal tools your team runs.
before dcouple, we built and ran a healthcare messaging product with 30m+ messages sent, six-figure arr, and hundreds of clinical customers. that taught us what software looks like when it has to earn trust inside real work; now we're building ai tools that carry that same trust into every knowledge workflow.
together, they point at the same thesis: as software becomes easier to create, the hard part shifts from using interfaces to trusting, sharing, and operating the work agents produce.
our products
pane. run any coding agent, on any os, from desktop or phone.
pane is not an ide, terminal emulator, or another agent. it is an open-source (agpl-3.0), keyboard-first surface for claude code, codex, aider, goose, or whatever cli ships next.
create a pane, get an isolated worktree, built-in diff viewer, browser tab, and git workflow. remote pane lets a vm, wsl box, desktop, or server run the repos and agents while you control them from another desktop or your phone.
live at runpane.com.
doozy — a to-do list where every task comes with an agent that completes it. you give it work through voice, text, or ambient meeting capture. it connects to 800+ apps (gmail, slack, notion, hubspot, linear, google drive, and more) and executes tasks autonomously with your approval.
three core primitives:
ears
passively captures meetings and voice memos without bots joining your calls. context from every conversation becomes actionable.
hands
connects to 800+ integrations and takes action: sends emails, updates crms, creates tickets, files documents, posts messages. everything requires human approval before execution.
brain
remembers how you work, your preferences, your context, and gets better over time. persistent memory across conversations means you never repeat yourself.
what it actually does, in the wild:
we connected bank exports and receipts, then used doozy to sort 1,200 transactions, draft ledgers, and assemble tax inputs for review across three states.
a sales leader had doozy read meeting transcripts against crm records, flag a deal that had never been entered, and draft the missing fields for review.
a user asked doozy for 34 follow-up drafts. it pulled crm notes, wrote account-specific first passes, and queued them for approval before anything went out.
a student described a job-search routine by voice. doozy turned it into a scheduled workflow that searches every morning and appends matching roles to a google sheet.
live at usedoozy.com.
morph. cloud workspaces for the internal software teams are already building with codex, claude code, or any terminal agent.
pane is for external software. morph is for internal software: import an existing github repo into a shared cloud workspace with auth, a database, and a collaborative vm where teammates can watch agent changes happen.
live at trymorph.dev.
the team
parsa khazaeepoul, co-founder. ex-ai2 incubator, where he supported ai startup founders with customer discovery and go-to-market. employee #1 at two computer vision / vlm startups, including head of developer relations at moondream ai (8k+ github stars). multi-hackathon winner (openai, yc tech week). organized 17+ turing and nobel laureate speakers through turing minds at uw. hosts the seattle consumer ai founder meetup series through tactical gtm. b.s. in informatics from the university of washington (summa cum laude).
tyler brown, co-founder & cto. yc w17 alum. previously built a game studio that reached 2m users and deployed to 1,600 applebee's locations. ceo of bloomtext, a hipaa-compliant clinical messaging platform — a $250k arr saas business he runs almost entirely using doozy, from customer support to tax filing across three jurisdictions. background in making complex technology radically simple for non-technical users.