Service

Agent Build

An end-to-end production AI agent built into your workflow. Three to four weeks. Fixed price. Connected to your real systems — not a demo.

Book a 25-min call
Fixed Price
£18,000–£28,000
Exact price agreed in Week 1 scoping
Duration
3–4 weeks
Plus 30-day post-launch performance review
Integrations
Up to 3 systems · Up to 2 data sources
Best for
Mid-market firms with a defined use case ready for production
What it is

Production AI — not a proof of concept

You've either run a pilot or you already know what you want built. You need an agent that runs in your actual environment, talks to your real systems, and handles the workflow end-to-end — not a demo that someone has to maintain, extend, and eventually rebuild before it's useful.

The Agent Build is a complete, production-ready system. We connect to up to three of your systems and two data sources, implement governance and logging from day one, run 40+ evals before we deploy, and measure your KPIs at launch and again 30 days later. You know it works, you know what improved, and you have documentation your team can use.

Deliverables

What you get

One production agent connected to up to three systems and two data sources
Evals harness with 40+ test cases across standard flows, edge cases, and failure modes
Governance and logging built in from day one
Deployment to production in your environment
KPI baseline measured before and after deployment
30-day post-launch performance review with metrics
Full handover documentation
Team training session (2 hours)

What you don't get

A multi-agent programme across your whole organisation. Ongoing engineering resource. A team embedded in your business. The Agent Build delivers one excellent, production-ready agent and the evidence to justify the next one. If you need an ongoing engineering team, that's the Agent Programme.

The Process

Four weeks to production

Week 1

Discovery & architecture

We align on the full scope — all systems, data sources, and integrations. We design the agent architecture, define the evals criteria, and agree the KPIs we'll measure. No surprises after this.

Weeks 2–3

Core build & integration

We build the agent, wire up all three system integrations, connect the data sources, and implement governance and logging. Daily async updates. You can review progress in our shared workspace at any point.

Late Week 3

Evals & hardening

We run 40+ test cases, covering standard flows, edge cases, adversarial inputs, and failure modes. Anything below bar gets fixed before deployment.

Week 4

Deployment & training

We deploy to production, run a 2-hour team training session, and hand over documentation. We measure your KPI baseline at this point.

Day 30

Performance review

We review the KPIs 30 days post-launch. You get a written report: what improved, what didn't, and recommended next steps. No charge for this — it's included.

This is for you if…

  • You have a validated use case (via a pilot or your own analysis)
  • You need a production system that handles real workloads
  • You're a mid-market firm that needs it to actually work, not a demo
  • Your leadership has approved budget for an AI initiative

This isn't for you if…

  • You haven't validated the use case yet — start with an Agent Pilot
  • You need multiple agents built and maintained — see the Agent Programme
  • You need an ongoing embedded team with retained capacity
Common Questions

Questions answered

What's the price range based on?

£18k covers one agent with up to three integrations into well-documented systems. £28k covers more complex scenarios: custom connectors, multi-step reasoning, legacy system integrations, or heavier governance requirements. We scope the exact price on the discovery call — there are no mid-project surprises.

What are KPIs and how are they measured?

We agree two or three measurable outcomes before we start — for example, 'average time to complete invoice processing', 'percentage of queries resolved without human escalation', or 'hours per week spent on [specific task]'. We measure the baseline before deployment and report against it at day 30.

Do we need existing cloud infrastructure?

You need an Azure subscription (or Microsoft 365 with Copilot Studio licences). If you don't have these, we can help you set them up as part of the project — typically adds 1–2 days and is included in the price.

What if we want to change scope mid-build?

We sign off scope at the end of Week 1. Changes after that are priced as change orders — clear scope, clear price, agreed before we proceed. We don't start and hope.

What's the difference between this and the Agent Pilot?

The Pilot is designed to test one use case quickly and cheaply. The Build is for teams that have already validated the use case (or don't need to) and want a production system that actually runs the workflow — multiple integrations, full governance, KPI tracking. Most clients do a Pilot first.

Agent Build

Ready to ship to production?

Tell us the use case. We'll scope the price and timeline on a 25-minute call.

£18,000–£28,000 fixed price · 3–4 weeks · KPI review included

Flux Assistant

Online

This assistant can make mistakes. Any pricing, costings, or financial figures mentioned are illustrative only — a Flux expert will provide accurate numbers for your project.

Hi, I'm the Flux assistant. Ask me anything about our services, pricing, or how we can help your business.