TempsTemps

Temps roadmapv0.1.0 · Stabilize

Build self-driving products.

Most platforms keep your product online. Temps is building the system that helps it improve itself.

First make the platform reliable. Then connect what users experience, what the code changed, and what the business needs next—turning that context into safe, explainable action.

See the pathWhat v0.1.0 stabilizes
System modelClosed-loop product operations
Stability first
  1. v0.1.0Stabilize
  2. v0.2.0Observe
  3. v0.3.0Propose
  4. v1.0.0Autopilot
  1. 01
    Sense

    Capture what users, code, and infrastructure are doing.

  2. 02
    Understand

    Turn connected signals into cause, context, and memory.

  3. 03
    Decide

    Choose the next move against a goal, budget, and policy.

  4. 04
    Act

    Ship, fix, rollback, scale, or experiment with permission.

  5. 05
    Learn

    Measure the outcome and make the next decision better.

The operating model

Autonomy is earned, not switched on.

Every release expands what Temps may do only after the previous mode is measurable, bounded, and trustworthy.

  1. 01
    Policy before action

    Environments, permissions, budgets, and blast radius define what the system may do.

  2. 02
    Evidence before confidence

    Findings and proposals carry their source signals, confidence, expected impact, and verification plan.

  3. 03
    Humans keep the wheel

    Autonomy is set per capability. Consequential changes remain reviewable, reversible, and interruptible.

The path

Stabilize. Observe. Propose. Improve.

One improvement kernel, four levels of authority. Reliability, growth, lifecycle, and operations all follow the same governed path. Target windows are directional and move if the stability or safety gates are not met.

Expected timeline

Targets, not deadlines
  1. v0.1.0Stabilize
  2. v0.2.0Observe
  3. v0.3.0Propose
  4. v1.0.0Autopilot
  1. 01
    v0.1.0Target · August 2026
    In progress

    Stabilize what already works.

    Before Temps earns more autonomy, the platform itself must be boringly reliable. v0.1.0 is focused on fixing bugs, tightening boundaries, and making every capability already in Temps predictable in production.

    • Release gates across deployments, observability, data, backups, email, and AI
    • Typed tools, explicit write confirmation, sandboxes, permissions, and audit history
    • Regression tests, recovery drills, and rollback paths for every critical workflow

    You operate the product. Temps makes the foundation dependable.

  2. 02
    v0.2.0Target · Q4 2026
    Next

    Observe where the product can improve.

    Temps connects analytics, traces, errors, performance, SEO, revenue, infrastructure, and deployment changes. It turns those signals into a prioritized, read-only stream of improvement opportunities.

    • Evidence-backed findings with impact, confidence, urgency, and affected users
    • Durable memory for releases, incidents, decisions, experiments, and outcomes
    • Specialized observers for reliability, product, growth, revenue, and operations

    You decide what to improve. Temps tells you where and why.

  3. 03
    v0.3.0Target · Q1 2027
    Next

    Propose the smallest safe change.

    A finding becomes a scoped proposal with evidence, expected impact, risk, and a verification plan. Temps prepares the artifact in a sandbox, but does not silently change production.

    • Code, configuration, content, SEO, lifecycle, outbound, and operational proposals
    • Branches, pull requests, preview environments, QA evidence, and impact forecasts
    • An approval bridge governed by policy, risk, budget, environment, and blast radius

    Temps prepares the change. You review and approve it.

  4. 04
    v1.0.0Target · H2 2027
    Vision

    Improve the product on autopilot.

    Give Temps a goal and guardrails. Trusted loops can observe, propose, verify, ship, measure, and learn end to end—while every action remains bounded, attributable, interruptible, and reversible.

    • Per-loop autonomy controls, capability budgets, human overrides, and a kill switch
    • Independent author and verifier agents, QA gauntlets, canaries, and automatic rollback
    • Outcome attribution that updates product memory and improves the next decision

    You govern goals, policies, budgets, and the level of autonomy.

v0.1.0 · Current focus

Make the existing platform boringly reliable.

Stabilizing40 capabilities in scope

These capabilities already exist. v0.1.0 is the hardening release that makes them a trustworthy substrate for every observation, proposal, and autonomous action that follows.

01 / Ship

Change the product

Loop roleAct

A reliable path from source code to production, with safe cutovers, previews, rollbacks, and fleet controls.

02 / Sense

See the product

Loop roleSense

Requests, traces, errors, sessions, and infrastructure health stay connected instead of being split across vendors.

03 / Remember

Keep operational memory

Loop roleRemember

The data layer lives beside the application, so health, backups, recovery, and product context share one home.

04 / Govern

Bound every action

Loop roleGovern

Identity, permissions, secrets, audit history, and network protections define what people and agents may do.

05 / Act

Turn context into action

Loop roleAct

Automation starts with product context and stays bounded by sandboxes, approvals, and visible run history.

Build it with us

The future operator is a product governor.

Tell us where your product still needs a human to connect the evidence, decision, action, and outcome. Those gaps are the roadmap.

Shape the roadmapRun v0.1.0