private alpha

Give your AI coding agents a workstation that never disappears.

TermRoam gives Claude Code, Codex, OpenCode, Kilo, and any terminal-based agent a persistent server workspace you can resume from laptop, tablet, or phone.

Persistent project workspaces for long-running sessions Resume from laptop, tablet, or phone Agents keep running on a server you control Code, logs, tests, commits stay together
Join the Alpha See How It Works
claude — running · checkout-flow

your-server:8443
PROJECTS
+New project
OPERATE
BUILD
alpha
No projects yet. Create one to get your first persistent session.

AI coding agents are powerful. Their workspaces are fragile.

You start an agent on a real task — then the laptop sleeps, the SSH tab dies, or the Wi-Fi drops, and you’re left wondering what survived.

Close the laptop

…and wonder what kept running.

Switch projects

…and lose the active context.

Open your phone

…and fight a terminal that was never designed for mobile.

Ask an agent

…then lose the task in a markdown file.

Agents were supposed to save time. Instead you’ve picked up a new job: babysitting terminals — reopening sessions, re-pasting context, scrolling for the log that mattered.

If an agent is expected to do real development work, it deserves a real workstation — not a disappearing laptop tab.

TermRoam gives every agent a persistent place to work.

You shouldn’t have to keep a laptop open just so an agent can finish its job.

TermRoam gives each project a persistent server workspace — terminals, files, sessions, logs, and agent workflows — resumable from any device. We run our own company’s development on it every day.

Before

  • Laptop must stay open
  • SSH sessions disappear on disconnect
  • Terminals and logs scattered across tabs
  • Agent work hard to supervise
  • Long tasks fragile
  • Phone access awkward
  • Context gets lost between sessions

After

  • Each project has a persistent home
  • Agents keep running after the laptop closes
  • Resume from laptop, tablet, or phone
  • Logs, files, terminals, and follow-ups stay together
  • Check progress, intervene, or continue later
  • Mobile terminal built for real use
  • Prompts turn into tested commits

Folders. Projects. Sessions. Everything in its place.

A workspace holds folders, folders hold projects, and every project holds its own terminal sessions — agents, shells, logs, and history side by side. Switching an agent between coding sessions is one tap; running many in parallel across projects is just… normal.

Start an agent. Close your laptop. Come back to the work.

Step 01

Create a workspace

Give each project a persistent server home with its repo, sessions, files, and logs together.

Step 02

Launch your agent

Run Claude Code, Codex, OpenCode, Kilo, or a plain shell in a live terminal session.

Step 03

Resume from anywhere

Check progress from laptop, tablet, or phone and return to the same session when you are ready.

What builders use TermRoam for

Queue feature development

Write the spec from your phone; an execution loop builds it, tests it, and commits. Review the diff when it suits you.

Progress checks from a phone

Scrollback, logs, and a live cursor from the couch or the airport.

Parallel agents across projects

Run a loop per project side by side — each keeps its own sessions, files, and history. Switch in one tap.

Sessions that survive you

Resume Claude Code or Codex mid-conversation without re-pasting context.

Tests that run while you’re away

Long suites finish whether or not your laptop came with you.

Overnight refactors

Kick off the big rename at 6pm, review the diff at 8am.

Your personal agent can chat. Now give it somewhere real to work.

Chat-first agents are great for reminders, research, inboxes, coordination, and lightweight automations. But when the task becomes “fix this bug,” “update this app,” “run the tests,” or “open a pull request,” the agent needs more than chat. It needs a repo, a shell, dependencies, logs, branches, approvals, and a place to keep working.

TermRoam gives coding agents that persistent place.

Works alongside the personal agent you already run.

Build it. Run it. Same box.

Your TermRoam server isn’t just where agents write code — it’s where the things they build keep living. The internal dashboard, the personal Getting Things Done app, the automation that runs your business: agents build them in a project, then the box hosts them, on ports you control, behind your own domain.

Development environment and production home — the same machine you already pay for.

Watch a spec become tested commits.

This is a real capability in the alpha today — recorded behavior, not a concept video.

The execution layer: specs go in, tested commits come out — around the clock.

Last night this product shipped its own checkout portal: 7 chained specs · 36 verified steps · 30+ commits · 997 tests green — one Execute click.

The overnight run, as it happened

23:12sell-p1 queued — one Execute click
23:58p1 clean · 5 commits · onboarding checklist live
01:41p2 clean · 10 commits · billing lifecycle + dunning
03:15p3 clean · 7 commits · team & account management
05:24chain complete · 7 specs · 997 tests green
07:30operator reviews the diffs, blesses, ships

Real run, real commits — timestamps from the night this page’s checkout portal shipped itself.

Learn the substrate: the TermRoam field guides

Interactive, product-neutral guides that teach the mental models behind AI coding agents — written for people graduating from AI assistants to Claude Code and Codex. Simulations you can click, not screenshots.

All 11 field guides →

Straight answers.

Agents

Which agents does TermRoam support?

Claude Code, Codex, OpenCode, Kilo — and anything else that runs in a terminal. A session is a real shell; if it runs on Linux, it runs here.

Control

Do I control the server?

Yes. Each account gets a dedicated virtual private server — a real Linux box, not a shared sandbox. Alpha builders who prefer their own hardware can self-host.

Mobile

Can I use my phone?

Yes — TermRoam is designed for real mobile use: live terminals, scrollback, file/image upload, and an action bar built for touch. Install it from Safari as a home-screen app.

Persistence

What happens to my sessions?

They live on the server, not in your browser. Close the tab, lose Wi-Fi, put the laptop to sleep — sessions keep running. Projects and files persist across everything, including reboots.

Editor

Is this replacing my integrated development environment?

No. Edit however you like. TermRoam is where long-running agent work lives; a built-in editor covers quick changes.

Personal agents

Is this replacing my personal agent?

No — they pair. Chat-first agents coordinate; TermRoam is where execution-heavy coding work actually runs.

Do-it-yourself terminal stack

How is this different from a do-it-yourself terminal stack?

A plain remote shell stack proves the workflow matters. TermRoam packages it: projects instead of socket names, a mobile terminal that works, file uploads, one-tap resume — no config to maintain.

vs Codespaces

How is this different from Codespaces?

Cloud dev environments optimize for editing in a browser-based integrated development environment on ephemeral containers. TermRoam optimizes for persistent, long-running terminal agents on a dedicated server you keep.

Export

Can I export everything?

Yes — it’s your box. You have full terminal access, so you can push to a git remote or copy files off at any time. If you cancel, the server stays online through the end of your paid period, and your data stays intact for a further 30 days before the server is reclaimed. Full billing & cancellation policy.

Availability

What’s available now vs planned?

Now: persistent workspaces, multi-agent sessions, mobile access, provisioning, billing. In alpha: execution queues and spec-to-commit loops (see Proof). Everything else lives in the roadmap below, honestly labeled.

Built inside a real business.

We’re a private-equity-backed team of founders operating a luxury hospitality group. Running that business — small team, high standards, no time for tooling that fights back — forced this product into existence.

TermRoam is the operating layer we built for ourselves, packaged so other founder teams can run theirs the same way.

“By founders, for founders. We ship what we already trust our business to.”

Less babysitting. More shipped work.

You’re not paying for a terminal interface. You’re paying for a dedicated always-on workstation for your agents: persistent compute, remote access, project organization, long-running sessions, and mobile continuity — with none of the do-it-yourself virtual private server setup.

Solo
A dedicated server, yours alone
$35/month
billed monthly
$32/month 1 month free billed annually — $384/year
2 vCPU · 4 GB RAM · 80 GB SSD
comfortably runs
  • ~2 agents working simultaneously
  • long-running sessions with logs and files together
  • unlimited projects & sessions
Studio
Maximum headroom for agent fleets
$130/month
billed monthly
$119/month 1 month free billed annually — $1,428/year
6 vCPU · 16 GB RAM · 320 GB SSD
comfortably runs
  • ~8–10 agents working simultaneously
  • heavy parallel loops & large builds
  • unlimited projects & sessions
Managed backups — scheduled off-server snapshots of your machine. Paid add-on, billed separately — not included in the plan price.
Estimates, not limits — it’s your machine. If the hardware can run it, you can run it.

Annual billing is pay for 11 months, get 12 — $385, $715, or $1,430 per year. Same hardware, same product.

Alpha cohort pricing — locked in while you stay subscribed. Hardware specs may improve; prices won’t.

Prefer to self-host? TermRoam installs on any Linux box — self-hosting stays possible for alpha builders. Mention it in your request.

Request private alpha access.

$ join --alpha

What alpha users get

  • Early access before public launch
  • Hands-on setup help from the team — we onboard you personally
  • Direct roadmap influence — your use case shapes what we build next

No card required. Limited seats. We prioritize builders actively running terminal-based agents on a virtual private server or private server.

Future roadmap: where TermRoam is headed

TermRoam starts as a persistent workstation for AI coding agents. Over time, we want it to become a broader cockpit for agent-driven work. These are roadmap ideas, not the reason to buy the alpha today.

Projects, files & persistent sessions In alpha

Project workspaces that keep terminals, files, notes, screenshots, logs, and agent sessions together.

Switch projects or open your phone without losing the active session or the files around it.

Projects · persistent sessions · files · notes · screenshots · mobile resume

Execution queues & spec-to-commit loops In alpha

Reviewed specs become queued execution loops that build, test, commit, and chain the next piece of work.

The alpha already runs these loops for real work; the roadmap is about hardening the controls, visibility, and review surface.

Spec review · queues · model cascade · test evidence · commits · auto-chain

Team workspaces & provisioning Planned

Central management for teams running one dedicated server per operator, with release channels and per-server setup.

Keep team servers aligned without giving everyone the same machine or the same secrets.

Not sold yet — there is no team plan at launch. Individual Solo, Builder, and Studio servers are the only plans you can buy today.

Provisioning · release channels · per-server usage · billing hooks · team admin

Usage and cost visibility In alpha

Daily visibility into token throughput, provider mix, run cost, active days, and audit trails.

See what each operator and agent run actually costs before usage turns into a surprise.

Token heatmap · cost rollups · provider split · active days · efficiency ratios · audit trails

Workflow canvas Planned

A self-hosted workflow builder for repeated agentic and deterministic automations.

For when repeated work graduates from asking an agent once to running the same workflow reliably.

Graphical canvas · agent, webhook, email, and Slack nodes · scheduled triggers · execution history

Chat-agent integrations Planned

Let chat-first personal agents hand coding work to TermRoam and receive results back without exposing the whole server.

Message your codebase or agents like coworkers while TermRoam keeps the private server boundary intact.

WhatsApp · Telegram · Slack · inbound email · phone shortcuts · thread routing

Notifications and approvals Planned

Approve an agent’s next step from a notification instead of returning to a terminal.

Loops pause for human decisions and resume the moment you respond from your lock screen.

Push notifications · phone shortcuts · in-app approval · lock screen · loop resume

Getting Things Done toolkit for agents Planned

A David Allen-inspired task system that gives agents a real operating workflow instead of dumping commitments into markdown files.

TermRoam can give agents inbox capture, next-action routing, waiting-for tracking, and reviewable task state.

Inbox capture · today view · next actions · waiting-for tracking · projects, areas, and routines

Long-term ideas Exploring

Go to Market outbound cockpit Exploring

A multi-channel outbound cockpit powered by the same execution layer.

Turns outbound into a workflow-backed system instead of spreadsheets, scripts, and inboxes.

Prospecting · campaigns · deliverability · inbox triage · analytics and sentiment

Company operating cockpit Exploring

A Google Workspace-aware operating home screen for what matters now.

Operator home screen for running the company, separate from Projects which is for building software.

Today view · calendar and email summaries · Getting Things Done priorities · scheduled agent runs

Broader founder workflows Exploring

The operating layer we run our own company on, hardened for customers workflow by workflow.

Agents compile your morning brief, surface approvals, and refresh dashboards so the first thing you see each day is actionable signal.

Morning digest · approval queue · revenue pulse · calendar brief · routine agents