Handles unstructured input
Invoices, contracts, forms, emails and scanned PDFs are read, understood and turned into structured records - the category of work that traditional RPA could never reach without a rigid template.
Operations
Retire the manual work between your systems
The expensive work is rarely inside a system. It is the person moving information between them - rekeying invoices, chasing approvals, reconciling two exports that never quite match. That work is now automatable, including the messy, unstructured parts that defeated older rule engines.
From$3,500
One mapped process automated end to end, with exception paths and a human review queue.
Indicative starting price. The fixed fee for your scope is quoted after the two-day diagnosis, before any build begins.
Start small
One workflow in a week
Fixed scope, fixed price, five working days
Pick the one repetitive job that costs you the most time. We scope it on a thirty-minute call, build it, and it is running in your systems within five working days of kickoff. If it is not running by day five, there is no invoice.
What you get
- A thirty-minute scoping call that ends with a one-page spec you approve
- One workflow built in n8n or Make, whichever fits who will own it
- An AI step where it earns its place: classification, extraction or a first draft
- Error handling and an alert to a person when something fails, not a silent log
- Under version control with a handover note, so it is yours and readable without us
What fits
- One trigger: a form, a webhook, an inbox, a schedule or a new record
- Up to two systems connected, both with an API or an existing connector
- A process that already exists and is agreed, so we automate rather than invent
What does not
- Anything that needs a model trained or fine-tuned
- A workflow whose rules nobody can write down yet
- Systems with no API and no connector
When it is live you can stop there. No retainer is attached. If you want more built, the second workflow is quoted from the same spec format, and the diagnose phase is where a wider roadmap starts.
What makes it work
Three things we insist on
Exceptions route to people
Automation that only works on clean cases creates a worse mess. Every workflow has an explicit confidence threshold and a human review queue, so ambiguity surfaces instead of propagating silently.
Auditable by design
Each step records its input, its decision and its reasoning. When finance or a regulator asks why something was approved, the answer is a record, not an archaeology project.
Today Person
- 1 Open the document
- 2 Read and interpret it
- 3 Retype into the system
- 4 Check it against the order
- 5 Chase whatever does not match
After Automated Human review
- 1 Read and extract
- 2 Validate against your systems
- 3 Post the clean ones
- 4 Queue the ambiguous ones
Capabilities
What is actually included
- 01
Document processing
Extraction, validation and posting for invoices, POs, claims, onboarding packs and compliance forms.
- 02
Cross-system orchestration
Workflows that span ERP, CRM, helpdesk and finance without a person as the integration layer.
- 03
Approval routing
Policy-aware routing with thresholds, delegation and automatic chasing of the person holding it up.
- 04
Reconciliation and exception handling
Matching across sources, with unmatched items explained rather than dumped in a spreadsheet.
- 05
Operational visibility
Throughput, touch rate and cycle time per workflow, so you can see what automation actually bought you.
In detail
What this covers, specifically
A category name is not a scope. These are the individual pieces of work inside this practice - take the two that apply to you and ignore the rest.
- 01
Process discovery
Mapping where time actually goes, step by step, with volume and handling time measured rather than estimated in a workshop.
- 02
Cross-system orchestration
Workflows that span ERP, CRM, helpdesk and finance without a person acting as the integration layer between them.
- 03
Approval routing
Policy-aware routing with thresholds, delegation and automatic chasing of whoever is holding it up.
- 04
Reconciliation and matching
Records matched across sources, with unmatched items explained and queued rather than dumped into a spreadsheet.
- 05
Exception handling
An explicit confidence threshold and a staffed review queue, so ambiguity surfaces to a person instead of propagating silently.
- 06
Data entry elimination
Removing the rekeying between two systems that both hold the same record for different reasons.
- 07
Legacy RPA modernisation
Replacing brittle screen-scraping bots that break whenever a form or page changes.
- 08
Scheduling and dispatch
Allocating jobs, shifts or vehicles against constraints that are genuinely too complex to solve by hand each morning.
- 09
Audit trail construction
Recording input, decision and reasoning at every step, so "why was this approved" is a query rather than an investigation.
What you receive
Concrete artefacts, not a slide deck
- Process map with volume, cycle time and cost per step
- Automated workflows with defined exception paths
- Human review queue and interface
- Audit log and compliance documentation
- Operational dashboard and alerting
01 Diagnose Days 1-2
02 Prove Week 1
03 Integrate Weeks 2-3
04 Operate Ongoing
Around three weeks end to end. That comes from scoping tightly to one workflow - not from skipping a phase. Each still ends in evidence you can check.
Free calculator
Run the numbers yourself first
Further reading
Written up in more detail
- Rate limits and API quotas across a workflow Why adding workers produces 429s, how to budget concurrency across every API a workflow touches, and the backoff that actually works.
- Scheduled or event-driven: choosing a trigger When polling on a schedule is the right answer despite being unfashionable, when webhooks are worth the delivery complexity, and the hybrid that most systems need.
- Webhooks that drop: delivery, duplicates and ordering Why webhook delivery is at-least-once and unordered, what that means for your workflows, and the three defences that make an endpoint reliable.
- Where the hours actually go: finding automatable work A method for locating the work AI can take off your team without a six-month process-mining exercise, including the four shapes that look automatable and are not.
Let's talk
Start with one painful process
Pick the workflow your team complains about most. We will map it, cost it, and tell you honestly whether automating it is worth doing - including when the answer is no.
Related practices