AmiCore
The Governed Container
AmiCore is the runtime container in which 3forge components are configured, executed, monitored and governed, so every use case inherits a common operating model instead of implementing its own.
Why a container
Governance as a first-class citizen
Operating systems and infrastructure platforms are unable to address governance inside a complex application:
- Which data an application may expose to a particular user
- How an AI agent may invoke a business procedure
- Which component is consuming a stream
- How an action is recorded for audit
- How files, ports, plugins and resources are managed across every component within the application
Implemented application by application, those decisions leave two systems on the same infrastructure following different security, monitoring and operational models.
AmiCore adds a layer of control inside the process itself: a shared runtime where the cross-cutting concerns are addressed once and applied consistently across every component, interface and application within it.
AmiCore
AmiCore provides a common runtime that lets you architect components for each use case, from simple ones to highly scaled deployments. It is a software container managing:
Platform & interfaces
- Deployment
- Data access
- Plugins
- REST
- MCP
- JDBC
InfoSec
- Role-based access
- Login & authentication
- File-system & port access
- Model activity monitoring
- Auditing
Development
- 100% unit test code coverage
- IDE integration
- Inline debugger
- Unified scripting language
Inside the AmiCore container, architects can shape their deployment using the following components:
Relays
- Reliable connectivity to real-time feeds
- Message transformation and routing
- Horizontal scaling of data ingest
Explore Relays
Centers
- Data virtualization, persistence, replication and publication
- In-memory and historical SQL DB
- CEP Engine
Explore Centers
Gateways
- Governed distribution of real-time subscriptions
- MapReduce for querying distributed data
Explore Gateways
Web Managers
- Horizontal scaling
- Centralized management of user profiles and dashboards
Explore Web Managers
Plug-in model
- AmiCore uses a plugin paradigm for I/O which delivers a common implementation of each interface governed by the same policy
- Agents enter through a defined interface, receive only the context they are entitled to, and leave an attributable record
- Governance is inherited from the runtime, not added to each application afterwards
Examples of available plugins
Relational & warehouse DBs
Oracle · PostgreSQL · SQL Server · DB2 · MySQL · Sybase · Snowflake · Vertica · Impala · Hive · SQLite
NoSQL, cache & in-memory
MongoDB · Couchbase · Redis · Ignite · Hazelcast · HBase · Chronicle Queue
Messaging & streaming
Kafka · Solace · TIBCO EMS / StreamBase · IBM MQ · AMPS · gRPC
Market data feeds
Bloomberg / B-PIPE · OneTick · QuantHouse · LSEG Data & Analytics (aka Refinitiv)
Cloud, identity & security
AWS KMS / SQS · Azure Vault · Okta · SAML · LDAP · Web Entitlements · SSH
Files, formats & tooling
Parquet · Avro · Excel · PDF · Python · Git · Perforce · Email
Governance
AmiCore is designed to meet the application runtime requirements of financial institutions by offering a flexible corporate security model that contains, audits and governs applications, including AI-generated ones, with the following capabilities:
3 reasons AmiCore fits financial institutions
Own environment
- Computers
- Premises
- Cloud
- Mobile
Data never leaves
the estate
Full runtime transparency
- Performance
- Resources
- Audit
Inside the container
Purpose-built components help you build your tailored infrastructure
Relays, Centers, Gateways, Web Servers, Web Managers and Web Balancers, each configured, executed and governed by the same runtime.
Comparing options
Benchmarking 3forge Enterprise vs alternative architectures
3forge Enterprise offers a compelling paradigm for scalable, governed architectures across initial deployment, project acceleration, and compound benefits. Architects often need to compare with alternative offerings, from building from Open source and in-house components, to cloud-based data platforms. Click on a score to read each rationale.
Your perception may vary based on existing estates or experiences. Contact us for an exploratory call to compare notes.
| Dimension | 3forge Enterprise | Cloud data platform | Assembled streaming stack | Custom in-house build |
|---|---|---|---|---|
| Real-time state and sub-second response |
3forge Enterprise · 5/5 Top mark. Current state is held in memory inside the same process as the application logic, so there is no network hop between a query and its data. |
Cloud data platform · 2/5 Freshness is bounded by ingestion and micro-batch scheduling, and every query is a round trip to a remote service. Caching narrows the gap without removing it. |
Assembled streaming stack · 4/5 Not higher because current state has to be materialized into a separate serving store. Latency then depends on that store and on keeping it consistent with the stream. |
Custom in-house build · 4/5 Not higher because the result rests entirely on the team’s design and sustained expertise. Nothing in the stack guarantees it. |
| Historical analytics at petabyte scale |
3forge Enterprise · 4/5 Not higher because the analytic tooling around the historical database is narrower than a warehouse’s, and compute does not scale elastically on demand. |
Cloud data platform · 5/5 Top mark. Elastic compute, mature query optimizers and the deepest analytic and machine-learning ecosystem of the four. |
Assembled streaming stack · 2/5 Not higher because a stream is not an archive. History needs a separate store and a second query path, neither of which the stack provides. |
Custom in-house build · 3/5 Not higher because scale comes from whichever database is chosen underneath, and the partitioning, tuning and lifecycle work stays with the team. |
| Governance enforced inside the runtime |
3forge Enterprise · 5/5 Top mark. Identity, entitlements and audit are container-level services, so every interface resolves to the same model instead of each application implementing its own. |
Cloud data platform · 3/5 Not higher because control stops at the data boundary. Once data reaches the application, the platform can no longer see or constrain what happens to it. |
Assembled streaming stack · 2/5 Not higher because each component ships its own security model. One institutional policy has to be reconstructed across them, and it drifts as they version independently. |
Custom in-house build · 2/5 Not higher because governance is implemented per application. Two systems on the same infrastructure end up with different entitlement and audit models. |
| Time to a governed production application |
3forge Enterprise · 5/5 Top mark. State, entitlements, interface, reporting, monitoring and deployment arrive with the runtime, leaving the business logic as the work. |
Cloud data platform · 2/5 Not higher because only the data layer arrives ready. The application, its interface and its controls remain a separate program of work. |
Assembled streaming stack · 2/5 Not higher because integration precedes the first application: the stack must be assembled, secured and made operable before anything ships. |
Custom in-house build · 1/5 Not higher because every recurring capability is built before the first business-specific one. It is the longest path to a governed release. |
| Moving parts to operate and keep aligned |
3forge Enterprise · 4/5 Not higher because a distributed deployment still runs several components across regions and environments, each needing capacity and failover planning. |
Cloud data platform · 3/5 Not higher because the managed platform is only one tier. BI, notebooks, orchestration and the application layer are operated and upgraded separately. |
Assembled streaming stack · 1/5 Not higher because it carries the most components, the most vendors and the most upgrade cycles, and every interface between them is the institution’s to maintain. |
Custom in-house build · 2/5 Not higher because the whole stack and its lifecycle sit with the team that wrote it, including the parts nobody wants to own three years later. |