AI workflow automation for manufacturers and service firms
AI Workflows on Open Models, Running in Your Own Accounts
Open-weight models cost far less to run than closed AI services, keep your data in accounts you control, and can be tested, pinned and audited. We map where your team's hours go and put one workflow live, with a person approving every action.
Fixed scope. Your accounts, your data, your code. Serving the Southwest from Phoenix.
Why open models
Four Reasons Open Models Win for the Back Office
- Lower Running CostOpen-weight models cost far less per task than frontier AI services, and there are no per-seat AI licenses.
- Your Data Stays YoursThe model runs in an account you own, not in a chat app your staff paste client data into.
- Governed by DefaultThe model version is pinned, so it never changes under you without a test and your sign-off.
- Transparent and PortableThe model weights are public, so the model can be tested, audited and replaced.
You Are About to Hire for This. Start Smaller.
Manual, Repetitive Work
Your team retypes, copies and chases the same things every week.
AI With No Rules
Staff are already pasting client data into chat tools, and nobody wrote the policy.
A Hire You Cannot Find
An AI automation hire costs about $140,000 a year in salary alone and takes months to recruit.
Median base pay across 31 verified buyer-side AI and automation postings, September 15 to October 30, 2026. How we counted.
Open models
Lower Cost, Your Data, Full Control
We build on open-weight models such as OpenAI's gpt-oss, Meta's Llama, Mistral and Qwen. Here is what that gets you.
Lower Running Cost
Open-weight models cost far less per task than frontier AI services, and there are no per-seat AI licenses. You pay for the work the workflow does, and we measure that cost on your real work before you build more.
Your Data Stays Yours
The model runs in an account you own, not in a chat app your staff paste client data into. Your data is not used to train anyone's model.
Governed by Default
The model version is pinned, so it never changes under you without a test and your sign-off. A named person approves every action that sends, pays, files or deletes, and every run is logged.
Transparent and Portable
The model weights are public, so the model can be tested, audited and replaced. You can see which model ran each step, and you are never locked to one AI vendor.
| Closed AI Services | Open Models With workrr | |
|---|---|---|
| Cost | Per-seat licenses or per-token prices set by the vendor | Low per-task cost, measured on your own work |
| Your data | Sent to the vendor's service | Processed in an account you own |
| Changes | The model updates on the vendor's schedule | Pinned version, changed only with your sign-off |
| What you can inspect | The model is a black box | Public weights, a log of which model ran each step |
| Lock-in | Tied to one vendor | Swap models without rebuilding the workflow |
If a step truly needs a frontier model, we tell you why, show you the cost, and use it only with your sign-off.
How it works
Two Phases, Two Readouts. One Workflow Running at the End.
Phase 1 · Measure
Find Where the Hours Go
- Step 1Kickoff, access and a written schedule
- Step 2Interviews with the people who do the work
- Step 3Score every repeat job for hours, risk and effort
- Step 4Readout. Pick the one workflow to build
Phase 2 · Build and prove
Put One Workflow Live
- Step 5Build it in your stack
- Step 6Run it on real work beside your team
- Step 7Readout, handover and fixed-price plan
Who it is for
Firms With Heavy Back-Office Work and No Time to Hire for AI
Owners and operations leaders at 20 to 500 person companies: manufacturers and distributors, law, accounting, insurance agencies, property management, wealth management, healthcare admin and other back offices with heavy manual work.
Works with what you already run: Microsoft 365 and Outlook, Google Workspace, Business Central, NetSuite, Epicor Kinetic, SAP Business One, QuickBooks, HubSpot, Salesforce, n8n, Zapier, Make.
Low risk
A Fixed Price, a Written Scope, and a Credit If You Build
Credit
The full assessment fee is credited toward any build signed within 60 days.
Clear Scope
One workflow live, written into the kickoff summary with the schedule. Anything extra is quoted before work starts.
You Own It
It runs in your accounts, under your licenses, with a recorded handover. Nothing to cancel, nothing locked to us.
Proof you can check
We Build in the Open and Publish Proof
Every artifact we certify is signed and logged in a public transparency log anyone can check. It is the same rule we bring to your workflows: every action approved, every run on record.
As of Nov 22, 2026, from the public workrr trust ledger.
Verify any certificate yourself: npx @workrr-trust/certify verify
Pricing
Start With One Workflow. Keep Going Only If It Pays.
Not sure yet? Start with the free workflow plan: a 30-minute call and a one-page written plan, with no obligation.
Assess
$3,500 fixed
One workflow live
Map the work, put one workflow live, get the plan.
Build
$7,500 to $25,000
Fixed per package of 2 to 5 workflows
The workflows from your plan, built in your accounts and handed over.
Run
From $2,500 / month
Monitoring, fixes and one change a month
$5,000 a month adds a new workflow every month.
No obligation. Many clients stop after the assessment with one live workflow and a plan.
Who Does the Work
Shawn Bure, founder. 25 years in technology and operations, from senior engineer to founder and CEO.
Shawn runs every assessment himself: the person on your plan call is the person who does the work.
FAQ
What Owners Ask Before They Book
What does the AI Workflow Assessment cost?
$3,500, fixed. Pay in full, or pay $1,750 to book and $1,750 when you pick the workflow to build. The full fee is credited toward a build signed within 60 days.
What is the free workflow plan?
A 30-minute call with Shawn about the work that eats your team's hours, followed by a one-page written plan: the workflow we would start with, what it would connect to, the risks we see, and whether the assessment is the right next step. It is free, it is yours to keep, and there is no obligation.
What happens after the assessment?
Nothing, unless you want more. You can stop with one live workflow and a plan. If you want to keep going, a build is $7,500 to $25,000 fixed per package of 2 to 5 workflows, and running it is from $2,500 a month. No lock-in.
Is a person in the loop?
Yes. Every action that sends, pays, files or deletes needs a named person on your team to approve it, and every run is logged.
Which AI models do you use?
Open-weight models by default, such as OpenAI's gpt-oss, Meta's Llama, Mistral and Qwen, running in an account you own. If a step truly needs a frontier model such as Claude or GPT, we tell you why, show you the cost, and use it only with your sign-off. Your data is not used to train models.
Why open models instead of ChatGPT or Copilot licenses?
Four reasons. They cost far less per task, with no per-seat licenses. Your data is processed in an account you own. The model version is pinned and every action is approved and logged, so nothing changes without your sign-off. And the weights are public, so the model can be tested, audited and swapped without rebuilding the workflow.
Who owns what you build?
You do. It runs in your accounts, under your licenses, and the code and configuration are handed over with a recorded walkthrough so your team can run it.
Pick One Workflow. We Will Put It Live on Open Models.
Start with a free one-page workflow plan. No obligation.