Example build Fits: small team or ops team
Supplier invoice intake and matching
Invoices arrive, get extracted, matched against open purchase orders, and filed — with only the exceptions reaching a human.
Stack
Cloud Run (Python) · Document AI / Azure Document Intelligence · Xero / QuickBooks / Business Central API · Postgres matching ledger · Slack or Teams for exceptions
Timeline
1 wk audit → 3 wks build → 2 wks supervised run
Outcome
Every invoice arrives structured and pre-matched. Humans only touch exceptions.
Read the build → Example build Fits: small team or ops team
Quote and proposal generation
An enquiry arrives and a branded, correctly priced quote goes out the same working day instead of two or three days later.
Stack
n8n or Cloud Run orchestration · CRM API (HubSpot / Pipedrive / Dynamics) · LLM for narrative sections only · Price rules in Postgres · PDF rendering service · E-sign integration
Timeline
1 wk audit → 3 wks build → 1 wk parallel run
Outcome
Quote turnaround goes from days to the same working day. Pricing errors drop because pricing stops being retyped. Every quote is logged.
Read the build → Example build Fits: ops team
Support inbox triage with drafted replies
A shared inbox gets sorted, deduplicated and routed before anyone opens it, with replies drafted and held for approval.
Stack
LangGraph or Google ADK agent on Cloud Run · pgvector over the help centre and past tickets · Zendesk / Freshdesk / Outlook Graph API · Evaluation harness on 200 labelled historical tickets · OpenTelemetry tracing
Timeline
1 wk evaluation set → 4 wks build → 2 wks supervised
Outcome
The queue is sorted before anyone opens it. Agents edit instead of compose. Escalation is a rule rather than a hope.
Read the build → Example build Fits: ops team
Weekly operations reporting pack
Four systems get reconciled into one model and one pack, with the numbers computed once instead of rebuilt by hand every Friday.
Stack
Scheduled Cloud Run job · BigQuery or Postgres warehouse · dbt-style transformations · Looker Studio or a rendered PDF · LLM used only for narrative commentary, never for the numbers
Timeline
1 wk audit → 2-3 wks build → 1 wk parallel run
Outcome
Friday afternoon comes back. The numbers stop being arguable because the definitions live in one versioned place.
Read the build → Example build Fits: solo / small team
Enquiry to CRM, enriched and followed up
Leads arriving across a web form, DMs and a phone number land in one place, get enriched, and are followed up without relying on memory.
Stack
n8n or Make · CRM API · Enrichment API · Calendar booking link · Transactional email · Google Sheets as the owner-facing view when that's genuinely what's used
Timeline
2-3 weeks end to end
Outcome
No enquiry is unacknowledged for more than a few minutes. Follow-up stops depending on the owner's memory.
Read the build → Example build Fits: small team or ops team
Client and employee onboarding orchestration
A signed contract triggers the entire setup — folders, permissions, channels, tasks, invoice — done identically every time.
Stack
Cloud Run or Power Automate · Google Workspace Admin SDK or Microsoft Graph · Slack/Teams API · Asana/Jira/Notion API · Accounting API · Terraform for the templated resources
Timeline
1 wk audit → 2-3 wks build
Outcome
Onboarding is identical every time and takes minutes. Permission mistakes stop being a recurring incident class.
Read the build → Example build Fits: ops team
Contract and renewal obligation watcher
Contracts get extracted into a structured register so renewal dates, notice periods and price escalators stop living in one person's head.
Stack
Document AI / Azure Document Intelligence · LLM extraction with a strict schema and confidence scoring · Postgres register · SharePoint or Drive connector · Calendar and email alerting · Full audit log of every extraction
Timeline
1 wk evaluation set → 3 wks build → 2 wks supervised
Outcome
No renewal is missed by accident. Obligations become queryable. Every extracted field traces back to the clause it came from.
Read the build → Example build Fits: small team or ops team
Ask-your-documents internal knowledge search
Years of proposals, reports and SOPs become searchable in prose, with citations and with permissions respected by construction.
Stack
Vertex AI Search or pgvector with embeddings on Cloud Run · Drive / SharePoint connectors with ACL propagation · Gemini or Claude for synthesis · Slack/Teams bot · Retrieval-quality evaluation set · Per-query cost telemetry
Timeline
1 wk scoping and evaluation set → 3-4 wks build → 2 wks pilot with one team
Outcome
Institutional memory becomes queryable. The 'who knows about X' tax falls. Permissions are respected by construction rather than by policy.
Read the build →