Capacity forecast
1–2 weeks
Translate product traffic and AI workflows into a defensible model-usage and cost range.
All services
Estma service / Model capacity
Model inference capacity
Plan and put in place annual model API capacity for an AI product, with usage forecasts, provider fit, quotas, cost controls and the integration work that makes the capacity usable.
Where this starts
A good fit when…
- A live or launching AI product needs a more predictable inference budget
- The workload combines generation, extraction, embeddings, ranking or summarisation
- A team is considering annual capacity but has not modelled real consumption
- Credits are available but quotas, monitoring or application integration are incomplete
The operating problem
The difficult part is rarely the headline technology.
The engagement focuses on the surrounding system: boundaries, evidence, permissions, exceptions, people and the decisions the implementation must support.
- 01
Capacity is estimated from user counts without modelling requests, tokens or retries
- 02
Model, prompt, retrieval and caching choices change the cost profile materially
- 03
Rate limits, project quotas and budget alerts do not match the product boundary
- 04
Procurement is separated from architecture, evaluation and release planning
What leaves the engagement
Concrete output, not advisory residue.
- 01Workload inventory and base, growth and peak consumption forecast
- 02Provider, model and embedding fit assessment
- 03Annual capacity plan and provider-aligned procurement support
- 04API and SDK setup, project boundaries, quotas and rate-limit controls
- 05Usage monitoring, token-cost optimisation and exception alerts
- 06Operating runbook, ownership map and renewal checkpoint
Ways to start
Choose the smallest engagement that resolves the next decision.
Provision and integrate
2–4 weeks
Set up the agreed capacity, application integration, project controls, monitoring and handover.
Annual capacity programme
12 months
Combine agreed inference capacity with usage review, quota management and optimisation through the term.
What Estma needs from your team.
- Expected product traffic and growth assumptions
- Representative prompts, documents or workflow traces
- Current model, API and application architecture
- A product and commercial owner for capacity decisions
Inside the work
See the decisions behind the service.
Representative engagements show how this capability connects to the product, workflow, controls, and people around it.
Service questions
What teams usually ask.
Is this only the purchase of model credits?
No. The useful scope connects commercial capacity to the workload, provider configuration, application integration, quotas, monitoring and cost controls. Capacity without those decisions is difficult to operate.
Which model providers can this cover?
The provider follows the use case, data constraints, model quality, regional availability and commercial terms. The final scope names the provider and what its credits, quotas and support actually include.
Start with context
Is this the work you need?
Describe the current system, the decision ahead and the constraint that is making progress difficult.