Prepared for your team · 2026

AI automation for wholesale distributors.

We take the manual work out of order entry, pricing and the back office.

Brightwork Digital · David Peyton · Graham Peyton, PhD

What we hear

Your order desk retypes what customers already typed.

  • 01Half my order desk’s day is retyping POs from email.
  • 02Quotes take a day or two when the counter is busy.
  • 03Our pricing lives in spreadsheets and in two people’s heads.
  • 04We bought the ERP add-on. Nobody uses it.

The four things distributors tell us most, paraphrased.

What’s changed

AI can now do the paperwork.
Most distributors haven’t started.

  • 280×

    cheaper to run the same AI, 2022 to 2024

    AI got good at documents, and cheap.

    Software can now reliably read a messy PO, a scanned invoice or a supplier price file and fill in the fields, and draft the order in your ERP for a person to approve.

    Cost: Stanford AI Index, 2025

  • 95%

    of AI pilots show no measurable return

    Pilots stall at the systems.

    They break where the AI has to fit into the systems and work a business already has. Run with an outside partner, they succeed about twice as often as internal builds.

    MIT NANDA, The GenAI Divide, 2025

  • 16%

    have it working broadly

    Distributors are behind.

    93% of distribution executives call AI a strategic priority. Only 16% have it working across more than one part of the business.

    Distribution Strategy Group, 2026

Meanwhile, your order desk runs on what a few people know. About 1 in 10 order clerks leave the occupation every year, and the knowledge leaves with them. BLS

The problem and what it costs

Your team is typing information that already exists.

  1. 01PO arrives by email, PDF or fax
  2. 02Retyped into your ERP
  3. 03Price and stock checked by hand
  4. 04“Where’s my order?” calls
  5. 05Vendor bills matched by hand
  6. 06Late payments chased

  • Workforce

    $70k a year

    What one order-desk employee costs, with benefits. Every hour spent retyping orders is part of that. BLS

  • Pricing

    +13% profit

    What a 1% price increase adds to operating profit, because distributors keep about 7 cents of each sales dollar. Census

  • Cash flow

    47 days

    How long distributors wait to be paid. Every day faster frees a day of sales in cash: $68k for a $25M distributor. SEC filings

Three examples · a $25M distributor

$83k a year and $342k of cash, for a $25M distributor.

Workforce

Order entry from email, PDF and fax

Orders arriving this way example
30 a day
Minutes saved on each example
× 8 min
An hour, with benefits, 250 days BLS
× $33

Worth every year$33k

Pricing

Finding and fixing price leaks

Sales not on contract prices example, 40%
$10M
Margin recovered McKinsey
× 0.5 pt

Worth every year$50k

Cash flow

Getting paid faster

Sales per day $25M ÷ 365
$68k
Paid faster, with reminders and matching example
× 5 days

Cash freed up, once$342k

Example figures are replaced with your real numbers in the audit. Mature pricing teams recover 0.5 to 1 point of margin a year (McKinsey). Build prices are quoted after the audit.

Example automation

An emailed PO becomes a draft order.

  1. PO arrives by email, PDF or fax.
  2. Read: customer, part numbers, quantities, dates.
  3. Check: your item numbers, contract price, stock, credit.
  4. Approve: a person sees the PO beside the draft and clicks.
  5. Into your ERP as a sales order, logged.
Example order, made-up customer.

Unusual orders go to a person with the reason highlighted. If anything fails, orders drop back to the manual queue.

Proof: four systems we’ve built

Documents in, software does the work, a person approves.

  • Estate agency · 50 agents

    One system for deals, commission and compliance.

    8,400 h

    a year back across 58 staff, modeled. Built in about four months.

  • Chip design · global semiconductor company

    AI agent teams that design and verify microchips.

    20×

    faster design turnaround.

  • Scribble · clinical notes

    Turns a patient visit into notes, letters and billing codes.

    0 bytes

    leave the practice: it runs on the office computer.

  • Maxflow · small-business marketing

    An AI marketing team that plans, posts and follows up.

    <60 sec

    to answer a new lead.

Also built A 5,500-member association’s data platform (3,355 records fixed, zero errors) · a 2,967-sermon archive, transcribed and indexed · Revéra Labs, a telehealth clinic and prescription-only store

How we work

Start small, prove it, then grow.

  1. 1

    Paid audit

    2 weeks

    Map the work, set a baseline, rank opportunities by dollars. You approve the deliverables.

  2. 2

    Build one job

    6 to 10 weeks · fixed price

    Target agreed in writing. Progress you can see every day.

  3. 3

    Run and improve

    Monthly

    Support starts once results show. Month to month.

Every change request is tracked until it’s live. Brightmark, our own tool

230 logged228 donemedian 13 hours

Guarantees

Miss the agreed target, and the build is on us.

  • Money back

    • The build is refunded if the agreed target isn’t hit within 90 days of go-live.

  • A named person

    • A US-based person, not a ticket queue.
    • A one-hour response if orders stop flowing.
    • Monitored around the clock.

  • Yours and private

    • NDA first.
    • Your code, your cloud, your data.
    • Never used to train AI.

Runs alongside your ERP · switch it off in one click · 30-day handover plan

Return on investment · a $25M distributor

Three automations or six: what you get back.

−$50k $0 $50k $100k $150k $200k SigningMonth 12Month 24Month 36 $188k six $76k three
Cumulative net return, after build and support.
3 automations6 automations
Value per year, estimated$75k$169k
We plan on 75% of that$56k$127k
Build, one time example$25k$50k in two phases
Support, monthly example$1.5k$2.5k
Pays for itself~12 months~13 months
3-year net return$76k $2 back for every $1$188k $2.45 back for every $1

Counted from signing. Each phase takes 3 months to build and 3 to 4 months to ramp up. Three: order entry, “where’s my order?” replies and collections. Six adds price leaks, quote drafting and vendor bill matching. Prices are examples; real prices are quoted after the audit.

Who you’d work with

You work with the two people who build it.

  • Portrait of David Peyton

    David Peyton

    Runs Brightwork Digital

    Has run a business, and the systems behind one.

    • Ran his own chiropractic practice in Johannesburg for 10 years, and lectured at the University of Johannesburg.
    • Ran IT for a multi-location healthcare group, including systems used by 3,200+ providers in a Henry Ford Health program.
    • Since 2023, about 40 systems that take manual work off businesses in healthcare, finance, real estate and the trades.
  • Portrait of Graham Peyton, PhD

    Graham Peyton, PhD

    Engineering and AI automation

    Builds automation where one mistake costs millions.

    • Architect of a multi-team AI agent system that designs and verifies microchips.
    • Builds design automation pipelines and the high-stakes verification around them, on highly sensitive customer data.
    • Software and infrastructure engineer. PhD, Imperial College London.

Next step

Start with a two-week paid audit.

You get

  • A map of your order-to-cash work
  • Hours and dollars per task
  • A ranked list of automations
  • A fixed-price plan for the first build

Price

A fixed fee, credited toward the build.

We need

Two hours with you, 30 minutes with each department lead, and sample POs and vendor bills.

  1. 1 Agreement today
  2. 2 Kickoff within a week
  3. 3 Findings two weeks later
hello@brightworkdigital.co

Appendix

Straight answers

Will this replace my people?
No. It takes the typing off their plate, so you grow without adding headcount.
Won’t our ERP vendor do this?
ERP AI works inside the ERP. The manual work is at the edges: email, PDFs, portals, spreadsheets.
What if it gets an order wrong?
Nothing goes out without a person approving it.
Is our data safe?
Your cloud, your code, read-only ERP access wherever possible, never used to train AI.
Why not buy an off-the-shelf tool?
If one fits, the audit will say so.
What if you go away?
You own everything, it’s documented, and support is month to month.

Appendix · engineering automation

A multi-team AI design system.

Engineer sets the goalAI agent teams designAutomatic verificationEngineer signs offReleased

Anything that fails verification goes back with the reason.

  • Teams of AI agents design and verify chips under engineer direction.
  • Pipelines from spec to sign-off, deployed and run in the cloud.
  • Verification built for work where one error is a multi-million-dollar mistake.

20× faster design turnaround

Estate agency · 50 agents

Compliance built into every deal.

Leads, attorney emails, documentsDeals, commission, trust moneyCompliance checksLetters, payslips, reports
  • Anti-money-laundering checks, risk rating and a five-year audit trail on every deal.
  • No commission paid without two approvals.
  • A support assistant on every screen, and one that runs 16 everyday jobs.

8,400 h a year across 58 staff, modeled4 months to build

Appendix · health systems

Where privacy and rules aren’t optional.

Keystone

Member database and emailChecked sync every 30 minStaff work queues

3,355 records fixed, zero errors.

Scribble

VisitNote, letters, codesChecked against insurer rulesClinician signs

Runs on the office computer.

Revéra Labs

QuestionnaireVideo consultE-prescriptionValid script only

South African privacy and health rules (POPIA, SAHPRA, HPCSA) built in.

Maxflow · marketing automation

An AI marketing team that learns what works.

  1. Plans the work
  2. Owner approves
  3. Publishes to 8 platforms
  4. Watches the results
  5. Learns what works
  • A private AI agent per business, with roles it can hire.
  • Earns more freedom only after 85% of its proposals are approved.
  • Ties leads and sales back to each post.