Anyone can call a model. What breaks in production is the layer around it — the provider that rate-limits at 4am, the schema that drifts, the permission nobody scoped. That layer is what we build.

Agents that finish work rather than demo it — reading the ticket, checking the system, drafting the reply, escalating where being wrong is expensive.
Delivered as a scoped system with an accuracy bar agreed in writing, a runbook your team can follow, and full source in your own repository from the first week.
Retrieval over your own documents
Evaluation suites and accuracy thresholds
Fine-tuning and open-weight deployment
Human-in-the-loop escalation

Infrastructure that survives a bad Tuesday. Autoscaling, multi-region failover and cost controls that stop a runaway job becoming a five-figure invoice.
Delivered as a scoped system with an accuracy bar agreed in writing, a runbook your team can follow, and full source in your own repository from the first week.
AWS · GCP · Azure · bare metal
Kubernetes and serverless GPU
Infrastructure as code, reviewed like code
Spend alerting and circuit breakers

Prompt injection, exfiltration through a model, over-permissioned agents — the attack surface AI adds is new, and most teams have not mapped it yet.
Delivered as a scoped system with an accuracy bar agreed in writing, a runbook your team can follow, and full source in your own repository from the first week.
AI-specific threat modelling
Secrets, key rotation and scoped access
Audit trails an auditor can read
PDPA and GDPR posture

A retrieval layer is only as honest as the index behind it. We build the pipelines, the versioning and the freshness checks that stop a model confidently quoting last quarter.
Delivered as a scoped system with an accuracy bar agreed in writing, a runbook your team can follow, and full source in your own repository from the first week.
Pipelines, warehousing and lineage
Vector stores and hybrid retrieval
Quality monitoring and drift alerts
Real-time and batch, same contract

The application around the model. Web, mobile and internal tools built by the same team that built the inference layer, so the seams are ours to answer for.
Delivered as a scoped system with an accuracy bar agreed in writing, a runbook your team can follow, and full source in your own repository from the first week.
Full-stack product engineering
API design and versioning
Design systems and front-end
Handover with a tested runbook

Our own tools, available to you directly — a routing gateway, an evaluation harness and a monitoring console we built because we needed them ourselves.
Delivered as a scoped system with an accuracy bar agreed in writing, a runbook your team can follow, and full source in your own repository from the first week.
Unified model gateway
Evaluation and regression harness
Cost attribution console
Self-host or managed
Ranges from engagements we have actually closed, published so a finance conversation can happen before an engineering one. The binding figure arrives at the end of the discovery week.
| Typical build | Indicative band | Monthly run | Main risk | |
|---|---|---|---|---|
| Artificial Intelligence | 5–8 weeks | USD 28k–60k | USD 3k–8k | No agreed definition of a correct answer |
| Cloud | 2–6 weeks | USD 14k–40k | USD 1k–4k | Migration scope creeping past the brief |
| Cybersecurity | 3–5 weeks | USD 12k–32k | Included in run | Findings nobody is resourced to fix |
| Data | 3–8 weeks | USD 15k–45k | USD 1k–4k | Turning into a full data programme |
| Software | 6–12 weeks | USD 30k–90k | USD 2k–6k | Requirements arriving after the build starts |
| Products | Same week | No build fee | USD 500–5k | Paying for a console nobody opens |
Bands assume one system, not a per-seat licence, and they exclude the model spend itself — that goes on your own provider account, so the number you see there is the number they charged. Platform pricing is separate and starts where the routing saves more than it costs.
Self-host if a payload leaving your network is a compliance problem, or if you already run Kubernetes and would rather own the uptime. Managed if you want the routing table maintained for you as provider prices move. The code is identical; only who runs it changes.
The request moves to the next entry in your fallback list mid-flight, and the caller sees a normal response with a different resolved model in the trace. You get an alert; your users get nothing unusual.
If every provider on a route is down, the gateway returns a structured error rather than a timeout, so your application can degrade on purpose instead of hanging.
Yes, and most teams do. Your keys stay in your secret manager and are used at request time. You keep the direct billing relationship and we never hold a credential that can spend your money.
Three signals, weighted by you: measured accuracy on your own evaluation set, p95 latency against the budget for that route, and cost per completed task including retries. Any model can be pinned by hand, and pinning is logged so nobody quietly overrides the data.
Yes — cloud, security, data and software engagements that stand on their own. We take roughly six to eight a year and turn down anything where we would be the third vendor in a room with no technical owner.