What it solves
Information arrives as email, forms and documents; someone reads it, decides what it is, re-types it and sends it on. That work is repetitive but not mindless — it needs judgement. We build workflows where an AI step handles the reading and classifying, rules handle the routing, and a person approves only the cases that genuinely need one.
Typical problems
- Enquiries answered late because nobody has triaged the inbox
- Data copied by hand from PDFs and emails into the CRM
- Drafts written from scratch when 80% is the same every time
- No record of who approved what, or why
- An AI pilot that impressed in a demo but was never wired into a real process
Example workflows
Three workflows we build regularly in this service. Names and numbers are illustrative of typical outcomes; measured results are published only in case studies.
Inbound enquiry triage
Every email is read by a person, forwarded to the right colleague and answered when someone gets round to it.
The model classifies intent and urgency, extracts the details, creates the CRM record and drafts a reply; a person approves the draft in one click.
Typical result: first response inside the hour instead of the next day.
Supplier invoice capture
Invoices arrive as PDF attachments and are keyed into the accounting system by hand, with occasional typos.
Attachments are extracted with confidence scores; anything under threshold goes to a review queue, the rest posts straight to the ledger.
Typical result: keying time removed; exceptions handled in one daily review.
Proposal first draft
A consultant writes each proposal from a previous one, copying and adapting sections.
Call notes and CRM data feed a template; the model drafts the bespoke sections, the consultant edits and sends.
Typical result: proposals out the same day as the call.
Capabilities
- Lead qualification and routing
- Document and email extraction with confidence scoring
- Drafting and summarising with human sign-off
- Classification, tagging and deduplication
- Approval steps with an audit trail
- Fallbacks and review queues when the model is unsure
- AI cost tracking per workflow
What every build includes
The difference between an automation that works in a demo and one that still works in a year.
- Automatic retries when a connected app is briefly unavailable, so nothing is silently lost
- A run log for every execution — what came in, what was decided, what went out
- Failure alerts to Slack or email after repeated errors, with a pause switch you control
- Idempotent steps: running twice never creates duplicates
- Written handover: a one-page map of the workflow, its owners and how to change it
- Built in your own accounts — you keep every credential, licence and asset
How the work is delivered
Map
Current process documented with the people who run it.
Scope
Written proposal, diagram, deliverables, fixed price.
Build
Short stages in your accounts, demo at the end of each.
Launch
Monitoring, alerts and documentation before go-live.
Pricing and engagement
We map one process end to end and tell you honestly which steps an AI model should handle, which need rules, and which still need a person.
Scoped in writing after the audit. One price, one outcome, no hourly surprises. Half on start, half on launch.
Monitoring, fixes when an app changes its API, and a monthly improvement sprint. Cancel any month.
Deliverables
- A working workflow in your own accounts
- Prompt and rule documentation, versioned
- A test set with expected outcomes so changes can be checked
- Monitoring dashboard and an error inbox
Questions
Which AI models do you use?
Whichever fits the task and your data policy — OpenAI, Anthropic Claude, Google or an open model on your own infrastructure. We keep the model behind an interface so it can be swapped later.
Is our data safe?
Data stays in your accounts. We use business-tier APIs that do not train on your data, minimise what is sent to the model, and can redact personal data before it leaves your systems.
How do you stop the AI making things up?
By giving it a narrow job, structured outputs, confidence thresholds and a human approval step wherever the cost of a mistake is real. The model reads and drafts; people decide.
Do we need a data science team?
No. These are workflows, not models. Your team needs to know the process; we handle the rest and hand over documentation.
How does pricing work?
Every engagement starts with a free audit. You then get a fixed-scope proposal with one price and a delivery date. Ongoing monitoring and improvements are a monthly retainer you can cancel any month.
How quickly can something be live?
A first workflow is usually live within one to three weeks of sign-off. Larger builds are staged so something useful ships early and the rest follows in sprints.
What happens when a workflow fails?
Every build ships with retries, a run log and alerts. If an app changes its API, the retainer covers the fix; without a retainer we quote a small fix and turn it around quickly.
Who owns the automations afterwards?
You do. Everything is built in your accounts and documented, and your team is walked through it at handover. You can maintain it yourself or keep us on retainer.