A REFERENCE ARCHITECTURE FOR COMPOSABLE AI

An open agentic
platform is a stack,
not a suite.

Build agentic systems from open data foundations, model choice, interchangeable harnesses, and portable standards, without surrendering the seams.

EXPLORE THE STACK ↓READ THE KNOWLEDGE BASE →
REFERENCE STACK / 01COMPOSABLE BY DESIGN
04Open standards5 OPEN COMPONENTS
03Harnesses & brokers8 OPEN COMPONENTS
02Models & routing5 OPEN COMPONENTS
01Data & semantics5 OPEN COMPONENTS
YOUR POLICIESYOUR INFRASTRUCTUREYOUR CONTROL

THE ANTHEM / VIDEO

Open the stack.
Keep the options.

A beat-synced visual manifesto for composable AI: open data foundations, model choice, interchangeable execution, and portable standards.

7 minutes 37 seconds · Press play with sound for the full manifesto.

DEFINITION / 00

Open components.
Explicit contracts.
Operational freedom.

An open agentic platform is an architecture in which data, models, execution, and interoperability remain independently understandable and replaceable. “Open” may describe source, weights, formats, or interfaces. A trustworthy architecture labels the difference instead of flattening it.

ARCHITECTURE / 01 TO 04

Four layers.
No mandatory vendor.

Each layer answers a different question. Together they turn model capability into durable, governable work. Every name below has a full explanation in the knowledge base. Names marked ↗ are example implementations Alex Merced builds; their full pages live on alexmercedai.com.

THE OPENNESS TEST / 06

Open is a property
of the whole system.

A pile of open-source parts can still produce a closed architecture. Test the relationships as carefully as the licenses.

01

Replaceable

Can one component be swapped without rebuilding the system?

02

Inspectable

Can a builder understand what runs and why?

03

Portable

Can identity, skills, context, and work move?

04

Bounded

Are authority and approval requirements explicit?

05

Grounded

Do agents share durable data and semantic meaning?

06

Auditable

Can people reconstruct decisions and outcomes?

SCORE YOUR STACK →Interactive, runs in your browser. Download the result as JSON or print it.

WORKED SCORECARDS

Score two possible stacks.

These are illustrative designs, not ratings of named vendors. Give each test 0 (absent), 1 (partial), or 2 (demonstrated). Record the evidence before buying or building.

TestComposable stackSingle-suite stack
Replaceable2: engine and model APIs are swappable0: proprietary orchestration couples both
Inspectable2: prompts, tools, and traces are exported1: console shows traces but no export
Portable2: profile and work use open schemas0: agent state is trapped in the suite
Bounded2: explicit tool scopes and approvals1: role permissions without per-action approval
Grounded2: governed metrics and versioned tables1: connected data without shared metrics
Auditable2: run, approval, and data versions recorded1: partial logs with short retention
Total12 / 124 / 12

Use the knowledge base to define each criterion, then score your own deployment with evidence in the interactive openness scorecard.

A PRACTICAL PATH / 06

Build from the ground up.

Start with durable context. Add intelligence and execution only after control boundaries are clear.

  1. 1
    Ground the system

    Choose open formats, a catalog, and a semantic layer that agents and people can share.

  2. 2
    Define the contracts

    Express identity, skills, tools, workflows, policy, and approval points in portable forms.

  3. 3
    Compose the runtime

    Select models, routers, brokers, and harnesses according to the work, not brand gravity.

  4. 4
    Observe and evolve

    Retain evidence, evaluate outcomes, and replace components as requirements change.

THE OPEN AI LIBRARY / 07

Read the systems
behind the stack.

Alex Merced has written 31 nonfiction books on AI, agents, semantic context, production architecture, and the data foundations beneath them.

Browse the complete book catalog ↗

Scroll to explore all 31 titles →

THE PRINCIPLE

Own the architecture.
Keep the options.

EXPLORE THE KNOWLEDGE BASE →

Work with Alex