Skip to content

Proof

Work we have delivered, described the way we would describe yours.

Client names are held in confidence, which is also how we will talk about you. The industries, the constraints and the numbers are real. We will walk any of these through in detail with your team.

Consumer products manufacturerSharp seasonal retail peaksHundreds of POs a day, three person deskLegacy ERP, not replaceable

We put the order desk's judgment into software.

The real problem

POs arrived as email, PDF and Excel, got printed, marked up by hand, then re-keyed into the ERP. The rules that make an order correct lived in three people's heads, not in any system. Order to ship took two to three days, and at peak the desk was the constraint on service.

Why it was hard

The ERP is old and load bearing, so replacing it was off the table. We had to work with the system they have and codify what nobody had written down.

A member of an order desk working at a computer in a warehouse office

What we built

  1. AI order intake.

    A secure cloud module sits in front of the shared order inbox and classifies every message, purchase order or not, product or service. Service POs route to their owner and never touch the ERP.

  2. Order extraction from email, PDF and Excel.

    Order data is pulled from the email body or its attachments, then assembled into a single clean order however the customer sends it.

  3. Seventy two rules, reviewed by the team.

    Business rules lifted out of people's heads and documented in the order they apply, with clear conditions for each, so execution is consistent and auditable. The team edits its own rules in a screen where AI flags conflicts and loops.

  4. Review and approve before posting.

    A review screen shows the assembled order for a person to approve or adjust. One click then writes it into the ERP, returns the order number and releases it to fulfillment.

The value here is the asset. A process that lived in three people's heads now runs as something the company owns, on the ERP it already had, and the order desk spends its day on exceptions and customers instead of typing.

10 to 20 hours
a week returned to the team
Same day
fulfillment, was 2 to 3 days
6 to 8 weeks
to measured impact
72 rules
documented out of people's heads into a company asset
Automotive aftermarket distributorUS arm of a global brandNo system of record, disconnected SaaSAggressive growth mandate

A new ERP, built around how the business should run.

The real problem

The US arm of a global brand, growing fast on disconnected applications, data and SaaS tools with no system of record to scale on. Growth is the mandate, and the operation as it stands today cannot absorb it. A new ERP is the answer.

Why it was hard

Migration, configuration and integrations eat an ERP budget and crowd out the question that actually matters, which is how the business should run.

An operator working on a production line

What we are building

  1. Process first, designed with the client.

    Our time goes into the future state operating model and the SOPs, built alongside the client. We stay at the level of how the business should work rather than disappear into configuration.

  2. The agentic stack does the build.

    Business specifications are translated into ERP configuration, data structures, screens, process implementations and automations. The specification becomes the system.

  3. A direct integration for the statement feed.

    A direct connection for the daily parent company statement feed brings the data into the ERP without the proposed file transfer middleware.

  4. Experienced practitioners carry the deployment.

    A small team of experienced business and AI practitioners runs the full implementation, so the client pays for business judgment rather than a bench of configurators.

The client spends its time deciding how the business should run, and gets a secure, scalable system built around that answer. We are tracking timeline, cost and team size against the conventional benchmark and will publish the measured numbers when the system goes live.

Targets we are tracking

Live
in under 12 months
Integrated
systems, data and insights
3 to 5x lower cost
than typical
Medical waste services companyGrowth aheadLean AR teamCollections chased by hand

An AR command center on their own data.

The real problem

Collections chased by hand across spreadsheets and systems, with growth ahead and a lean AR team. Who to call first, what is at risk and why, and what it is all worth lived in a few people's judgment rather than in one view.

Why it was hard

The signal was spread across billing, the bank, the CRM and email. The value case had to be agreed against a baseline before anything was built, so the return could be measured rather than asserted.

A logistics team moving through a warehouse

What we are building

  1. Baseline first.

    Days sales outstanding, aged receivables, cash collected per week and hours spent chasing, pulled and agreed before build.

  2. One view of every account.

    Forecast, risk, priority and next action in one screen, built on the company's own data structures.

  3. Reminders, pre checks and invoice validation.

    The routine work carried by software, with a person deciding on the accounts that need a conversation.

  4. Measured against the baseline.

    Cash freed from receivables and hours returned, reported week by week against the numbers agreed at the start.

The company grows without the AR work growing with it. The team spends its day on the accounts that need judgment, and the number at the end is one we agreed to be measured on.

Value case agreed at the start

2.3x to 6.8x
return in year one
Cash freed
from receivables
Growth
without adding AR headcount

Next

Tell us where work gets held up.

Tell us the workflow that costs you the most hours in an ordinary week, or the work you keep turning down because there is no capacity to chase it. Thirty minutes is usually enough to know whether there is something here.

Start the conversation