Whether you are commissioning a voice agent, a finance AI workflow, or a knowledge agent, the engagement shape is the same. Discovery week to scope and validate. Build engagement to ship the agent. Managed retainer to keep it running and improving. We do not over-engineer the process; we apply it consistently.
One week. Fixed scope.
We map your processes, your tools, your data, your compliance environment. Pick the first agent shape. Design the integrations and the approval gates. Specify what the agent will do and what it will not do.
You leave the discovery week with three things: a written specification, a build estimate, and an honest read on whether you should buy a vendor product or commission a custom build. We will tell you to buy if buying is the right answer; the discovery fee covers our time either way.
4 to 16 weeks, depending on scope.
The agent gets built end to end. Voice infrastructure, RAG layer, integrations, compliance, audit trail, deployment, monitoring. We deploy in your tenant (your Azure, AWS, or self-hosted), with your model provider keys, so your data does not feed any vendor's training.
Goes live with human-in-the-loop on every commit until your team trusts it. Trust grows as you watch the agent work, not as we tell you to. We hand over runbooks, documentation, and ownership at the end of the build.
Optional monthly retainer.
We monitor accuracy, push fixes when output drifts, retrain as your business and policies evolve, run quarterly reviews against the metric the agent was built to move. Model and infrastructure costs passed through at cost.
You can run the agent yourself if you have the team. The retainer is for clients who would rather we handle the operational layer so the internal team can focus on what to build next.
We do not commit to your ledger autonomously. Every action that hits your finance, CRM, or production system goes through an approval gate. The agent does the work; the human signs off on the consequence.
We do not deploy on vendor-hosted models that train on your prompts. Production AI builds in client tenants run on Azure OpenAI, AWS Bedrock, or self-hosted models. Your data does not feed anyone's training set.
We do not skip discovery. Builds without discovery are usually rebuilds. The discovery week saves five weeks of wasted build time. Anyone offering to skip it is selling you something.
We do not run on offshore teams. Named delivery, UK-based studio, work hours that overlap with US East and West and Australia. Your build is run by the people you spoke to in the discovery call.
We do not lock you in. The agent is yours. The code is yours. The configs are yours. If we part ways next year, the agent keeps running. No per-seat trap, no proprietary platform you cannot leave.
We are based in London. Discovery and build engagements happen by video call, with optional on-site days for clients within Europe. The work itself is async; we ship updates with documentation, and we meet to make decisions that need a real-time conversation.
UK and EU clients get UK office hours. US East and West clients get morning and afternoon overlap depending on the side of the country. Australian clients get the early UK morning slot or the late UK evening slot, both workable on a per-week-not-per-day cadence.
Pricing in GBP for UK, USD for US, AUD for Australian engagements. Net 30 invoicing for UK and EU, Net 14 for US and Australian engagements unless agreed otherwise.
