Job
Answer from your actual policies
Not the internet's version. Retrieval over your documents with citations, so the answer is auditable and anyone on the team can see where it came from.
Eastbourne · UK
/ AI Solutions / AI AssistantsAssistants that know how your business actually runs.
Internal copilots grounded in your systems, your documents and your permissions — not just the public internet. Built for the jobs your team actually does, inside the tools they already use.
01/ What this actually is
Not a chatbot. A colleague who's read the file.
Most people's first AI assistant is Copilot or ChatGPT answering from the internet. That's a useful start, and occasionally the whole job. The work we do begins where that stops — an assistant grounded in your actual policies, your CRM, your matter files and your inboxes, so the answer it gives is specific to your business and traceable back to the source.
02/ Jobs it does
Six shapes of assistant work.
Narrow, well-scoped jobs an assistant can take off a human's plate — reliably enough that the human stops checking every answer.
Job
Draft with your tone and your data
Proposals, replies, file notes, handover briefs — prefilled from the CRM or PSA record, written in a voice that reads like your team, not a model.
Job
Summarise the long stuff
Email threads, matter files, tender responses, call transcripts — reduced to the shape the next person actually needs, not a generic bullet list.
Job
Look things up across systems
One question, answered by pulling across CRM, PSA, SharePoint, the shared drive and the inbox — instead of the user tabbing through five apps themselves.
Job
Trigger the next step
An assistant that can open a ticket, draft an invoice, book a slot, send the follow-up — not one that stops at talking about doing it.
Job
Hand off with context
When the assistant hits its limit, the human inherits the full thread, the intent and what's already been tried — not a restart from scratch.
03/ How we'd approach it
One job, grounded properly, in the tool people already use.
Assistants that try to do everything end up trusted with nothing. We'd rather ship one that earns its keep and grow from there.
Pick one job, not a platform
Assistants fail when they try to be everything. Start with one well-defined job where the user would notice if it went away tomorrow.
Ground it in the right data
The documents, the systems, the inboxes that actually hold the answers — connected properly, with retrieval tuned for how your people really ask.
Wire in permissions from day one
The assistant sees what the user is allowed to see. Identity and access controls are part of the build, not a compliance review six months in.
Put it in the tool people already use
Teams, Slack, a browser tab, an email thread — wherever the work already happens. Not a new app with a new login to remember.
For the governance side — permissions, grounding, audit and the identity perimeter that keeps an assistant honest — see the AI security perimeter.
04/ Copilot vs custom
Honest about which is actually the right answer.
Plenty of teams have a Copilot licence they're not getting much out of. Sometimes the fix is configuration, not a new build.
Sometimes the right answer is configuring the Copilot licence you already pay for, properly — the grounding, the prompts, the permissions, the rollout. Sometimes it's a custom assistant reaching into systems Copilot can't. We start every engagement by working out which, and saying so — even when it turns out to be the smaller piece of work.
05/ Where it lands
Useful in the tool people already have open.
An assistant that lives in its own app gets forgotten. The ones that stick are the ones sitting inside the workflow someone is already in.
/ Into your systems
The assistant reads from, and writes back to, the systems that already run the work — CRM, PSA, ERP, SharePoint, the document store. That integration work is our day job, documented on the systems integration page.
/ Into your workflows
When the assistant triggers a next step, it lands in the automation plumbing the rest of the team already relies on — tickets, queues, approvals, notifications — handled the way our workflow automation work is handled.
/ Backed by
Delivered by M-Tech Labs with the compliance and security discipline of M-Tech Systems — Cyber Essentials certified, aligned to NCSC CAF 4.0 and progressing through the Assurix trustmark programme. Permissions, retention and prompt-logging handled as first-class concerns, with code continuously scanned for quality and security via Aikidoand hosting kept on continuously pen-tested, current-version infrastructure.
Back to AI Solutions/ Start a conversation
Start with one assistant, one job.
A short discovery call on where a grounded assistant would earn its keep first — and whether your existing Copilot licence gets you most of the way before anyone writes new code.