Supertamer
See pricing

The agent work system for solo product builders

Let agents work. Keep the product in your hands.

Supertamer turns durable intent into agent work without losing context, decisions, or control. Tasks, memory, sessions, and automations stay connected in one native Mac app.

Purchasing opens soonExplore the system

Built for macOS 14+ · Apple Silicon

Supertamer5 agent sessions active
Product concept · Based on the current beta model
Agents working03
Needs you01
Designed forNative macOS
Source of truthLocal-first
Personal syncPrivate iCloud
Agent choiceBring your own

01 · The new bottleneck

More agents should not mean more management.

Parallel agents create a new kind of drag: stale instructions, scattered sessions, and decisions that disappear between runs.

01You re-explain the same product context.
02You poll every session just to find the one that needs you.
03Fast execution outruns the product’s source of truth.
Speed is not the bottleneck. Continuity is.

02 · One quiet control plane

See the work. Preserve the truth. Shape what runs next.

Supertamer holds the durable layer between agent runs. It shows what changed, what is true now, and where your judgment is actually needed.

Interactive product concept · Interface may change during beta
Work graphWhat can move nowAutomation presetHow that work should move
WORK GRAPH

The graph answers what should happen next.

Task hierarchy and prerequisites share one view, so you can see what is ready and why other work is blocked.

  1. 01
    Capture the work as a task

    Write the goal, definition of done, and anything that must finish first.

  2. 02
    Choose a ready node

    Start ready work now and send independent tasks out in parallel.

  3. 03
    Open a session from the node

    The agent starts with a brief built from current canon and earlier results.

  4. 04
    Bring the result back

    Done, waiting, and needs-you states immediately reshape what can run next.

Ready to startWaiting on earlier workNeeds your judgment
CUSTOM AUTOMATIONS

Describe how you work once. Run it on any board.

Choose Create preset and describe selection, parallelism, review, and stop rules in plain language. An agent turns them into a reusable automation preset.

Type this in the composer
Pick only ready release tasks, run up to three in parallel, verify every result, and pause when product judgment is needed.
✦ Preset createdSafe release rounds
  1. 01
    Choose Create preset

    Open the same composer you use for a session from the + beside Automations.

  2. 02
    Describe the strategy

    Say what to select, how much to parallelize, and when to review or stop.

  3. 03
    Name the target at run time

    Reuse the preset and add this run’s board or task scope in plain language.

  4. 04
    Run in finite rounds

    Every execution ends at a declared boundary and leaves a verifiable hand-off.

Next time, one sentence

Run the Release readiness board with ‘Safe release rounds.’

If judgment is needed, it pauses and appears in Feed.

03 · Durable intent

Store what should last. Plan against what is true now.

Tasks hold the goal, reason, constraints, and definition of done. The implementation plan is generated at execution time from current code, decisions, and dependencies.

Persist intent. Generate plans just in time. Execute against the current truth.
Task #T1842Ready
What

Recover active requests when a session token expires.

Why

A token refresh currently drops the builder’s in-flight work.

Constraints

Renewal belongs to the networking layer. Never duplicate refreshes.

Done

Expiry → refresh → retry works under concurrent requests.

Execution planGenerated from current truth when an agent claims the task

04 · Memory that compounds

Every run leaves the next one better prepared.

Agents do not just finish tasks. They write decisions and results back to the product’s living memory, so the next session can retrieve only what matters.

01IntentA task names the change
02ExecutionAn agent works now
03DecisionWhat changed is recorded
04CanonCurrent truth is updated
05Next briefRelevant context returns

05 · Your agents, one lasting system

Use the agents and hosts you already trust.

Supertamer owns the work between executions—not the model, credentials, or inner agent loop. Change the harness without losing the record.

01

Built-in agent sessions

Connect compatible agents directly and follow the conversation inside Supertamer.

Local · interactive
02

Paseo sessions

Launch, monitor, and rejoin local or remote sessions through one continuous surface.

Local + remote
03

External hosts

Let an agent in your preferred host read and update products, boards, tasks, threads, and canon.

Stable references · local tools

06 · A different category

Not another tracker. Not another agent launcher.

Supertamer is the persistent product layer that stays when tasks, sessions, hosts, and models change.

What survives?Task trackerAgent launcherSupertamer
Durable work intentPartialRun promptNative
Cross-session product memoryManualHost-boundCanon + ledger
Live intervention surfaceNoSession onlyFeed + graph
Choice of agent hostExternalUsually fixedBuilt in

07 · One product, yours to keep

Buy the tool. Keep the work.

No account-priced tiers and no subscription required to keep using the version you bought. One license covers the full Supertamer system.

The product page is public; purchasing is not open yet
Permanent licenseMac · Apple Silicon
$79one-time

The complete app and one year of product updates.

Updates after year one$39 / year

Renew for another year of updates. If you do not renew, your last eligible version keeps working.

  • All product, memory, agent, and automation features
  • Private iCloud sync across your Macs
  • Your data remains readable and exportable
Purchasing opens soon

Supertamer for Mac

Scale agent work without scaling your supervision.

Build with more continuity, less polling, and a product memory that gets stronger every time work is done.

Purchasing opens soonNative macOS · Local-first · Private iCloud sync