---
title: "The platform beneath the products."
canonical_url: https://auteurintelligence.com/platform/overview/
description: "Every Auteur product sits on one shared platform. It is what lets a single project move from a manuscript to a breakdown to a finished shot without inventing…"
---

# The platform beneath the products.

Every Auteur product sits on one shared platform. It is what lets a single
project move from a manuscript to a breakdown to a finished shot without
inventing a new identity system at each step, and it is why the work can survive
a change of AI harness.

This page is the map. Four platform pages go deeper:

- **[Algernon Engine](https://auteurintelligence.com/platform/algernon-engine/)**, the runtime and authority
  layer every product is powered by.
- **[Auteur Conductor](https://auteurintelligence.com/platform/conductor/)**, the harness control surface
  that lets the AI client you already use operate authorized Auteur domains.
- **[Project Intelligence](https://auteurintelligence.com/platform/project-intelligence/)**, the persistent
  memory, canon, and evidence that outlive the chat window.
- **[Local, Hybrid & Private](https://auteurintelligence.com/platform/local-hybrid-private/)**, where the
  work runs, stated honestly, including what is not yet live.

---

## One engine, carried across every workflow.

Auteur products are powered by one persistent engine that carries project
intelligence and governed execution across workflows. Orchestration, memory,
canon, evaluation, policy, receipts, application control, and execution are not
re-implemented per product, they are the shared substrate underneath all of
them, which is how an approved decision made while writing is still an approved
decision when the same project reaches production.

---

## Work moves through a durable lifecycle.

Complex goals move through a durable, reviewable workflow instead of
disappearing into a chat session.

Approved truth stays human-reviewed and versioned, and approved canon stays
separate from automated suggestions, so broad automated extraction never becomes
canon on its own.

---

## Many agents, distinct roles.

Auteur can assign different models and roles to planning, making, testing,
critique, and final judgment. Those distinct roles support reviewable
deliberation under shared budgets and bounded context, and durable decisions are
preserved rather than lost when a session ends.

The tools and project context an agent receives are grounded and auditable:
agents get the tools and context relevant to the task instead of an
unrestricted, drifting transcript, drawn from a hash-validated operations
authority.

---

## Verification is independent of self-report.

Auteur verifies what changed and what passed instead of accepting an agent's
claim that the work is complete. It derives Git ancestry, commits, changed
paths, diff validity, tests, and mismatches independently of the executor's own
report, the record is built from what happened, not from what was claimed.

---

## Built to be operated by the harness you already use.

Auteur Conductor is a harness-agnostic control surface that lets a compatible
external AI client discover and operate authorized Auteur domains in natural
language. Connecting an interface never grants wildcard authority; authority is
recomputed on every call, so a tool left visible in a stale client list cannot
act beyond what is currently authorized.

---

## Where it runs is stated plainly.

Owner-local operation runs the engine and your work on your own machine. A
mutually authenticated local bridge, which keeps proprietary media on your own
machine while a hosted control plane coordinates approved work, is designed and
substantially tested but its live persistent transport is not yet complete. Private, project-scoped hosted
deployment is designed for tenant isolation but carries no production hosted
claim yet. The [deployment page](https://auteurintelligence.com/platform/local-hybrid-private/) gives the full
truth table, including what is not live.