AI coding cockpit

Stop babysitting your AI coding agent.

Not another AI IDE — the transparency layer around the tools you already use.

Tellumetry shows what your agent is doing, what it will cost, where it may fail, and when you actually need to come back.

Sits around the tools you already use

  • Claude Code
  • Cursor
  • Warp
  • GitHub
  • Vercel
What it is

A calm command center for AI-assisted coding.

Tellumetry turns AI coding from a black-box terminal session into a transparent, ambient command center. It sits beside your agent and answers the four questions a scrolling terminal never will.

01

Estimate the cost

Know the token and dollar spend before a task runs — not after the bill arrives.

02

Map the progress

Watch the agent move through Repo Scan, Plan, Edit, Test, Deploy, and Review.

03

Preview the risk

See where a run is likely to stall, break, or touch something it shouldn’t.

04

Reclaim your attention

Step away. Tellumetry pings you only when a decision genuinely needs a human.

The real problem

Agents work fast. Watching them is exhausting.

AI coding agents can run for minutes or hours, but you can’t look away. There’s no cost meter, no map of where they are, no warning before things go sideways. So you sit and watch a scrolling terminal — burning your attention to babysit a tool that was supposed to give it back.

  • You watch a scrolling terminal because you don’t trust it to run unattended.
  • You have no idea what a task will cost until the tokens are already spent.
  • When a run goes wrong, you find out three steps too late.
  • You can’t tell the difference between “thinking” and “stuck.”
Interactive demo

Run a preflight check before you hand off the task.

Configure a task, model intensity, risk tolerance, and how you want to be notified. The Agent Preflight Simulator estimates tokens, cost, risk, and files touched, then animates the agent pipeline end to end. Mock data only — nothing connects to your repo.

Live previewRuns in your browserMock data

Preflight

default model mix

ConservativeNormalAggressive

Mock preflight only — no agent runs and nothing connects to a repo.

Run a preflight check before handing work to an agent.

Pick a task, set the model intensity, risk tolerance, and notifications — then simulate the run to see cost, risk, and the files it would touch.

Core features

Everything the terminal hides, on one calm surface.

Tellumetry doesn’t run your code. It reads the run and translates it into cost, progress, risk, and attention.

  • Cost estimation

    Per-task token and dollar estimates before the agent starts spending — and a live tally as it goes.

  • Pipeline map

    Repo Scan → Plan → Edit → Test → Deploy → Review, rendered live and legible instead of buried in logs.

  • Risk preview

    Flags risky edits, missing tests, and likely failure points before the agent commits to them.

  • Attention routing

    Ambient notifications that fire only when a human decision is actually required — and stay quiet otherwise.

  • Token load

    Real-time token burn and context pressure across the run, so you see the squeeze before it stalls.

  • Tool-agnostic layer

    Sits over Claude Code, Cursor, Warp, GitHub, and Vercel — a transparency layer, not a replacement.

Example use cases

For anyone who lets an agent off the leash.

  • Solo builders

    Ship without hovering

    Kick off a task and do real work. The cockpit watches the run and taps you only when it matters.

  • Teams

    Shared visibility

    Everyone sees the same cost, progress, and risk on one surface — no more “what is the agent doing?”

  • Agencies & freelancers

    Predictable cost

    Estimate token spend per task before you start, so quotes and budgets survive contact with reality.

  • Long migrations

    Unattended runs

    Start a big refactor or migration and trust the attention checks to bring you back at the right moments.

Why this is different

Tellumetry doesn’t replace your agent. It makes it legible.

Most AI coding tools want to be the place you write code. Tellumetry wants to be the place you understand it — the instrument panel around whatever agent you already trust.

Not an IDE

We don’t replace your editor or your agent. We wrap them in a layer you can actually read.

Transparency, not autopilot

You stay in control. We just remove the blindfold and show you the run as it happens.

Cost-aware by default

Every task carries an estimate before it runs, so spend is a decision, not a surprise.

Attention is the product

The goal isn’t more dashboards to stare at. It’s giving your focus back until it’s needed.

The Ailiur ecosystem

Part of the Ailiur context ecosystem.

Tellumetry shares the same idea as the rest of Ailiur: software that understands your context. Your projects, your tools, and your working style travel with you across the ecosystem.

Early access

Get Tellumetry before your next big run.

Tellumetry is in private development. Join the waitlist to help shape the cockpit and get early access as we open it up.

No spam — one note when your cohort opens.

Feedback

Give us your honest feedback.

Tellumetry is early, and it's shaped by the people who use it. Tell us what's working, what's missing, and what we should build next — it goes straight to the team.