CapabilitiesCapability

Shipping it was the easy part

Serving, observability, cost control and reliability for AI systems that are already in production and now have to stay correct, affordable and defensible.

The part nobody budgets for

An AI feature reaching production is the start of its cost curve, not the end. Content drifts away from what the system was tuned on. Providers change model behaviour and pricing underneath you. Traffic patterns shift. Someone discovers a prompt that makes it misbehave. None of this shows up in a launch plan, and all of it lands on whoever is on call.

01Serving, scaling and failover across model providers
02Quality monitoring, not just uptime monitoring
03Spend attribution per feature, per customer, per workflow
04Drift detection against a fixed evaluation set
05Incident response for the failure modes that are specific to AI systems
Where the money goes

Why AI bills surprise people

Cost optimisation is the most common reason we get called about an existing system.

01

Nobody can attribute the spend

One API key serves eight features, so the invoice is a single number with no breakdown. You cannot tell which feature is expensive, which customer is unprofitable, or which change last month tripled the bill. Attribution has to be instrumented at the call site before any optimisation is meaningful.

02

Everything runs on the largest model available

The frontier model gets chosen during the prototype and never revisited. A large share of real traffic is routine and would be handled identically by a far cheaper model. Routing by difficulty, with the expensive model reserved for the cases that need it, is usually the single largest saving available.

03

The context window is filled because it is there

Stuffing twenty retrieved passages into every prompt costs money on every call and frequently makes answers worse, because the relevant passage gets buried. Tighter retrieval is both cheaper and more accurate, which is a rare combination.

04

Caching was never set up

Repeated system prompts, repeated document context and genuinely repeated questions all get paid for again every time. Prompt caching and a semantic cache on common questions are unglamorous and immediately effective.

05

Retries are invisible and unbounded

A failing call that silently retries three times costs four times as much and appears in no dashboard. Retry budgets, timeouts and circuit breakers are basic reliability engineering that AI stacks routinely skip.

The engineering

What we put in place

Model routing by difficulty

Cheap models handle the routine majority, expensive models handle the hard minority, with the routing decision measured against the evaluation set rather than guessed. Quality is held constant while cost falls.

Routing, tiered inference

Spend attribution and budgets

Cost tagged per feature, per tenant and per workflow at the call site, with ceilings and alerts, so a runaway loop is caught in hours rather than at the end of the billing month.

FinOps for LLM workloads

Quality monitoring, not just uptime

A system can be fully available and completely wrong. Sampled outputs scored continuously against the evaluation set, with alerts on grounding and abstention rates rather than only on error rates and latency.

Drift detection, online evaluation

Provider abstraction and failover

No single provider becomes an unremovable dependency. Failover across providers, and the ability to move a workload when pricing, behaviour or terms change, which they do.

Multi provider, graceful degradation

Tracing across the whole call path

Full traces covering retrieval, tool calls, prompt version and model version, so an incident can actually be investigated rather than guessed at from a transcript.

Observability, tracing, replay

Governance you can hand to a reviewer

Model inventory, versioned prompts, change history, evaluation results over time and a documented human oversight path. What an internal audit, a security review or a regulator will ask for, prepared before they ask.

AI governance, audit readiness
Adjacent depth

The platform work underneath

AI systems sit on ordinary infrastructure, and most of the reliability problems we are asked to solve are ordinary infrastructure problems with an AI system on top.

Cloud and platform

  • AWS
  • Azure
  • Google Cloud
  • Kubernetes
  • Terraform
  • Serverless

Reliability

  • SLOs and error budgets
  • Incident response
  • Load and failure testing
  • Autoscaling
  • Disaster recovery

Cost

  • Cloud cost optimisation
  • LLM spend attribution
  • Rightsizing
  • Commitment planning
  • Egress reduction

Security

  • Secrets management
  • Least privilege access
  • Network isolation
  • Dependency scanning
  • Audit logging

Already live and hard to reason about

We take on other people's AI systems. A thirty minute call to understand what you are running, what it costs and where it hurts.