01
Workflow mapping
The task worth automating, the data it needs, who approves what, and clear acceptance criteria before any build work starts.
AI and automation · AI Systems & Automation
AI that runs inside your actual workflow
Most AI projects stop at a demo. We build agents and workflows that read your enquiries, draft your replies and move work between your systems, with a person approving what matters.
Starting from £5,000AI and messaging usage costs are agreed in writing before launch.
What changes
- Answers routine questions from your own documents, with sources shown
- Drafts replies, summaries and records for a person to approve
- Moves enquiries and information between your tools without retyping
- Flags anything it cannot handle to a named person instead of guessing
The gap
The problem worth solving.
The admin behind every enquiry, quote and handover still happens by hand in most established businesses: information retyped between systems, replies drafted from scratch, follow-up that depends on someone remembering. None of it needs judgement, but doing it manually is what costs the time. A general-purpose chatbot does not know your process, your documents or who is meant to sign off, so it never closes that gap on its own.
Our approach
We start with the workflow that costs your team the most time, not with a list of AI features. That means agreeing the data it can use, the tools it connects to and exactly where a person reviews before anything ships. The system is tested against real examples from your business before it goes near a live enquiry, and it is scoped to one connected system rather than a pile of disconnected experiments.
What is included
What an implementation covers.
Starting from £5,000. AI and messaging usage costs are agreed in writing before launch.
02
Agent or workflow design
A knowledge assistant, bounded agent or connected automation, scoped to the tools, permissions and rules your team already operates under.
03
Model selection
Built with OpenAI, Anthropic or Google models chosen per task on quality, integration needs and running cost, not one default.
04
Tool and system connections
Integrations into your inbox, CRM, calendar and documents, built with n8n or a direct connection where that fits better.
05
Review points and limits
Defined boundaries on what the system may send or decide on its own, and where it must stop and hand over to a person.
06
Testing on real examples
The system is run against genuine enquiries and edge cases from your business before anyone outside the project sees it.
07
Launch and handover
A controlled rollout with the people who will use it day to day, plus documentation of what it does and does not do.
Process
How the work runs.
Discover
We map the workflow, the data involved and where the time actually goes today, and agree what success looks like.
Design
We define the system architecture: models, data sources, tool connections and the exact points where a person reviews.
Build & connect
We build the agent or workflow and connect it to your live tools, testing continuously against real examples.
Launch & handover
We roll it out with your team, document how it works and hand over a system you understand and can run.
Ongoing
Optional ongoing support.
Monitoring, fixes and further optimisation once the system is live can be quoted separately once we know how it is being used.
Where ongoing optimisation, advertising management, SEO, support or AI operations are needed, we quote a retainer separately based on what the work requires. No implementation fee includes one by default.
What we measure
We record a baseline before launch, then track completion quality, time spent preparing and reviewing each task, and how often the system needs to ask for help. The comparison is always the real workflow against its own starting point, not a demo against nothing.
Stack
Technology this typically involves.
- OpenAIWe build with itDrafting, classification and summarising
- Anthropic (Claude)We build with itReading and reasoning over long documents
- Google (Gemini)We build with itMultimodal work and answers checked against search
- Custom AI agentsWe build with itBounded agents built around one business process
- n8nWe build with itConnecting tools into one workflow
Good questions
Before we talk.
What can you actually build?
Agents that answer from your documents, draft replies and update records within limits you set, and workflows that move work between your inbox, CRM and calendar. We agree the exact scope on the first call.
Will it make decisions for us?
No. People keep every decision that carries risk or judgement. The system prepares, drafts, routes and flags; a person approves.
What is not included?
Ongoing monitoring and further optimisation after launch, and any system outside the workflow we agreed to scope. Both can be added, but neither is assumed.
How is this priced?
Every proposal starts from a positioning price and is fixed after we have scoped your workflow. Ongoing support is optional and quoted separately once the system is live.
Which AI models do you use?
We build with OpenAI, Anthropic and Google models and choose per task on quality, integration needs and running cost. No single provider is the automatic answer.
Connected
Works with.
AI tools
A category guide to AI tools for recruitment companies
A category-by-category look at AI tools recruitment companies actually use: drafting, call notes, personal assistants, CRM features and custom agents.
AI for business
ChatGPT for business and what to build instead
What ChatGPT is genuinely good for at work, the data settings to check first, where it stops, and when a custom system is the better next step.
AI for business
How to build an AI workforce
What an AI workforce actually means: personal assistants, process agents and connected automations, how to choose between them and roll them out.
Your next step
See what this looks like in your business.
Starting from £5,000, scoped to your requirements. All prices exclude VAT.
No slides, no obligation.