Factorial SystemsFactorial.
Services · Four disciplines

We design, build, integrate, and operate — the full implementation arc.

Most engagements touch at least two of the four disciplines below. We scope them as a combined shipping plan, not a menu of line items.

/01

Conversational AI.

From customer-facing assistants to internal agent copilots — built on ChatCraft or from scratch where the fit demands it.

What we deliver

  • Solution design with conversation flow and handoff map
  • Retrieval layer over your data — vector, keyword, or hybrid
  • Guardrails, PII handling, and compliance review packaging
  • Channel wiring — web, voice, WhatsApp, or internal Slack
  • Analytics dashboards and evaluation harnesses

Stack signatures

ChatCraftLangChainLangChain4jOllamaPineconepgvectorAnthropic · OpenAI

Typical engagement

6–10 weeks for a scoped production deployment, including guardrail review. Most clients continue with us on a managed-ops retainer afterwards.

/02

Workflow automation.

Process orchestration that replaces manual handoffs — document review, approvals, data reconciliation, operational reporting.

What we deliver

  • Process mapping and AI-in-the-loop design
  • Event-driven services on Spring Boot, FastAPI, or Node
  • Document pipelines — extraction, classification, routing
  • Human-in-the-loop approval surfaces
  • Observability and audit trail instrumentation

Stack signatures

Spring BootFastAPIKafkaTemporalPostgreSQLMinIOKeycloak

Typical engagement

8–16 weeks depending on process complexity and integration count. We start with the noisiest handoff in your ops org and build outward from there.

/03

Integration engineering.

Wiring AI and automation into the systems you already run — core banking, ERP, CRM, legacy mainframes, NIBSS, and public-sector APIs.

What we deliver

  • Integration architecture with versioning and contract testing
  • API gateway patterns and tenant isolation
  • Auth flows — OAuth2, JWT, Keycloak, SSO
  • Event streams, webhooks, and data sync
  • Deployment on-prem, hybrid, or in your cloud of choice

Stack signatures

APISIXSpring SecurityOAuth2 · JWTKafkagRPC · RESTNIBSS · BVNGKE · AWS

Typical engagement

Scoped per system. We've integrated with NIBSS, core banking platforms, Odoo, ERPNext, Frappe, and several government identity services.

/04

Platform operations.

We don't deploy and disappear. Managed hosting, monitoring, tuning, and ongoing optimization for production AI systems.

What we deliver

  • SRE-grade monitoring, alerts, and incident response
  • Model evaluation harnesses and prompt regression testing
  • Cost and latency observability across providers
  • Quarterly architecture reviews with engineering leadership
  • On-call rotation covering West and East African business hours

Stack signatures

Prometheus · GrafanaOpenTelemetrySentryDocker · KubernetesTerraformLangSmith

Typical engagement

Monthly retainer. Scales with the surface area of what's in production. Most clients start here three months after a delivery engagement ends.

Engagement model

How an engagement actually runs.

/011–2 weeks

Discovery

Workflow mapping, data access review, success metrics, and an honest buy-vs-build assessment. Ends with a scoped proposal.

/022–3 weeks

Design

Solution architecture, integration contracts, security and compliance review, operational cost model. Deliverable is an ADR-backed plan.

/036–16 weeks

Build

Iterative delivery against fortnightly milestones. Your stakeholders are in the standups. No surprises at UAT.

/04Ongoing

Operate

Managed ops, incident response, and quarterly tuning. We operate it until you're ready to take it fully in-house.

Have a system you need
to actually get live?

Tell us what you're building. We'll tell you honestly whether it's a one-week ChatCraft deployment or a six-month platform build — or neither.