Deployed for a client
Meeting intelligence
The recorder captures the call, n8n collects the transcript, the model writes the summary and the task list, tasks go to the CRM and the summary to a channel. Post-call processing down 80–90%.
Service · AI automation
AI automation pilots with a before-and-after number
Call notes into tasks. Proposals from transcripts with your pricing logic. Weekly reporting from the data you already have. Each process goes live in 1–4 weeks, in production, and is measured against the number it started from.
PriceFrom £4,000 per process
Time to live1–4 weeks
Stackn8n · MCP · OpenAI / Anthropic / Google · Supabase
Runs onOur cloud or your infrastructure
01What we automate
Processes we have put into production
Deployed for a client
Proposal generation
Inputs from the chat or transcript, your pricing logic, a branded proposal. Preparation time down 70–90% for a payments company, pricing standardised.
Deployed for a client
Internal process automation
Data sync, event triggers and automatic internal communications. Manual workload down 60–70%, 8–12 hours a week saved per employee.
Deployed for a client
Call analytics
Transcription, speaker recognition, summaries and topics in batches, available in a dashboard and Telegram. Call analysis time down 80–90%.
Runs on our company
Candidate processing
An application form, a structured profile from the model, a card in the hiring channel and a record in the knowledge base. Ready profiles instantly, no manual work.
Deployed for a client
Content engine
Monitors sources, extracts insights, drafts posts and short-video ideas; a pipeline from research to delivery. Content production time down 70–80%.
Client names withheld under NDA. Figures are measured before and after on each project.
02How we work
From the matrix to production
Pick from the matrix
The audit ranks candidates by effect and effort. We take the top 2–3.
Measure before
Current time per task, volumes, people involved. Without a baseline there is nothing to prove later.
Build and launch
1–4 weeks per process, in production, with the people who run the process testing it. AI where it is needed; plain integrations where it is not.
Measure after, hand over
The before-and-after figures in writing, documentation, and the option of ongoing support.
FAQ
Questions we get asked
Which processes do you automate first?
The top of the opportunity matrix from the audit: usually meeting notes into tasks, proposals from call transcripts, weekly reporting and candidate processing. Each goes live in 1–4 weeks.
Do you always use AI?
No. If a process is deterministic, we automate it with plain integrations — n8n, Zapier or Make. A language model only adds cost and unpredictability where a rule will do.
Which tools do you build with?
n8n (including self-hosted), MCP servers, OpenAI, Anthropic and Google models, Zoom and tl;dv APIs for calls, Supabase and PostgreSQL for data, Telegram and Slack bots, Directus for custom panels.
How is a pilot measured?
We record the current time and volume before building, then measure the same figures after launch. You get the before-and-after numbers in writing.
What does it cost?
From £4,000 per process. The exact figure depends on the integrations and data involved, which is why we quote after the audit.
Have a process in mind?
Tell us what it is and how long it takes today. We’ll say whether it is a 1-week pilot or a 4-week one — and what it should return.