AI models
Claude automation, built around how you work.
Anthropic's models, used through the business API for reading, classifying, extracting and drafting.
- Tool
- anthropic.com
- Usually paired with
- n8n, Airtable, Supabase, Zendesk
- Timeline
- Most builds live in 3 to 5 business days from scope
- Owned by
- You. Accounts, credentials and code in your name.
What we build on Claude
- Add AI steps to workflows where the input is unstructured
- Extract fields from invoices, emails and documents
- Classify tickets and leads with a confidence score
- Draft replies for a person to approve
What we sync
Workflows we are asked for
- Invoice extraction with Claude into accounting
- Support ticket triage with Claude
- Lead qualification against an ideal customer profile
- Document summarising into a CRM note
Often paired with
Tools Claude usually needs to talk to.
-
Claude and n8n
The automation platform we build most systems on: code where needed, visual where not, cloud or self-hosted.
-
Claude and Airtable
A relational database with a spreadsheet face, used as the operations base for teams who have outgrown sheets.
-
Claude and Supabase
Postgres with auth, storage and APIs, and the database under most apps we build or rescue.
-
Claude and Zendesk
The helpdesk most support teams of 10 to 1,000 run on.
In every build
The same standards, whichever tool.
Under the hood
- Upserts keyed on the record
Every write is matched on the source ID. A retry or a re-fire updates the existing record instead of creating a second one.
- Retries with backoff
Rate limits and timeouts are retried with increasing delays, then parked for review, never dropped.
- Idempotent, loop-safe writes
Two-way flows carry a marker so a change written by the sync does not re-trigger the sync.
- Reconciliation
A nightly job compares counts and totals on both sides and lists any record that differs.
- Audit log
Every run records input, mapping result, action taken and who or what triggered it. Searchable by record.
For the team
- Slack or email notifications
The alerts go to a named person, not a channel nobody reads. You choose Slack, email, WhatsApp or all three.
- A runbook
One document per integration: what it does, what each alert means, and what to do in the first ten minutes.
- Training
A live session with the people who will use the review queue, recorded for whoever joins later.
- Handover
Admin access, billing and code sit in accounts in your name from day one. If we leave, it keeps running.
- Sixty days of hypercare
The engineers who built it stay on call after go-live, fix what breaks and tune what the team asks for.
Keys and data
- Keys live in a vault, never in code
API keys and tokens are stored in the orchestration layer's encrypted credential store or the platform's secret manager. None are written into workflows, repos, spreadsheets or chat.
- One credential per connection
Each integration gets its own key, scoped to the objects and actions it needs. A lead sync cannot read invoices. Revoking one key stops one thing.
- Service accounts, not personal logins
Connections use a dedicated service user in your account, so nothing breaks when a person leaves and nothing runs under someone's name.
- Two-factor on every account
Every platform account we set up has two-factor authentication on, with recovery codes handed to you, not kept by us.
- Rotation on handover and on request
Keys are rotated when the build is handed over, when anyone with access leaves, and at any time you ask.
The offer
Put Claude to work with the rest of your stack.
A fixed scope agreed in a thirty-minute call, live in three to five business days for most builds, accounts in your name, and the engineers who built it on call for sixty days after.
- ScopeThirty minutes to agree the workflows, the fields and the owner. Written up the same day.
- BuildValidation, upserts, retries, reconciliation and an audit log, on n8n, Make or code.
- Go-liveSandbox first, then a switch. Longer only where a vendor approval sits on the path.
- HandoverRunbook, training session, alerts to a named owner, admin access and code in your accounts.
- AfterSixty days of hypercare, then a care plan or run it yourself.
Questions
What people ask about Claude.
Is our data used for training?
Not through the business API. We use it with a data processing agreement, and only the fields the step needs are sent.
Claude or OpenAI?
Per task. We test both on your real data for accuracy and cost and pick per step. Switching later is a config change.
Claude
Start with the one Claude workflow your team still runs by hand.
Bring it to a thirty-minute call. You leave with a yes or no on fit, the shape of the build, and what we would do first.
Book the call