AI automation + products — Puget Sound, WA + remote

We don’t start with AI.

We start by watching how your business actually runs. The AI and automation come in only where it pays.

We build automation systems around real operating workflows.

More than twenty-five years in technology Microsoft product, field & applied research Mighty AI, acquired by Uber
Field note

The work nobody documented

Most operations accumulate a hidden layer of workarounds as tools get added, teams adapt, and the connections between systems never quite get built.

Retyped data The same record entered into multiple systems because none of them quite talks to the others.
Rebuilt reports The Monday report assembled by hand even though most of the information already exists somewhere else.
Tribal-knowledge handoffs The process works because one person remembers what happens next — until that person is out.

You can’t automate a workflow you haven’t understood. That’s why we map the work before we recommend the technology.

How we work

Three steps, in order. Because automation only works when it starts with the work.

01
M

Map

Know where the work actually goes.

We sit with the people whose work touches the process and build a workflow register — each task, how often, how long, and what breaks. Not a survey. Observation.

02
T

Triage

Know what is worth fixing.

Every workflow gets scored on volume, cost, and automation feasibility given the systems you’re locked into. You get a ranked shortlist with dollars attached — and an honest list of what to leave alone.

03
B

Build & measure

Prove the economics before scaling.

We build one or two workflows first and measure them against the baseline before anything else rolls out. If the numbers don’t hold, we say so.

You own the output at every stage. If you stop after the map, you still leave with an actionable operating picture — even if we build nothing further.

What we build

Built around the process, not the company.

01 — Diagnostic

Operations Diagnostic

A structured read on where your team’s time and money actually go, and what is worth automating. Yours to keep, whether or not you work with us again.

Best when You know there is friction, but you do not yet know which problem deserves the first dollar.
You’ll typically walk away with:
  • Workflow registers for every role in scope
  • Automation triage scored on cost and feasibility
  • A ranked shortlist — and an honest list of what to leave alone
  • ROI ranges with the assumptions shown, not hidden
02 — Pilot

Pilot Build

One or two workflows from the diagnostic, built and deployed so you can judge the result before committing to more.

Best when The opportunity is clear and you want evidence from your own operation before scaling.
The pilot is built to answer:
  • Does the workflow work in the systems you already use?
  • Does it improve the baseline we measured?
  • Do the economics still hold after real-world use?
  • Should we scale, adjust, or stop?
03 — Ongoing

Managed Intelligence

Ongoing operation of what we build — monitoring, upkeep, and adjustment as your systems and workflows change.

Best when A working automation is now part of the operation and somebody needs to keep it healthy.
Ongoing attention can include:
  • Monitoring deployed workflows
  • Upkeep as source systems change
  • Adjustments when performance drifts
  • Regular checks that the value still holds

Diagnostics are fixed-fee and scoped to the process, not the company. Ask and we’ll size it in one conversation.

Questions
We already pay for software that’s supposed to do this.

Usually true, and usually part of the problem. The diagnostic often finds capability you’re already licensed for and nobody turned on. We’ll tell you that before we quote you anything new.

Is our data safe?

We design around your constraints, including agreements that limit where your data can go. Where a workflow can run on infrastructure you own, we’ll build it that way.

How long before we see anything?

The diagnostic itself is a deliverable, not a prelude to one. The first pilot comes after that and is measured against the baseline — you’ll know whether it worked, not just whether it’s live.

What if the answer is that we don’t need AI?

Then we say so and you keep the map. That’s happened, and it’s a better outcome than an automation nobody uses.

From the build

One hour changed a product build.

~70%
of accounts where frontline staff could not identify who had last worked the customer

While building an insurance operations platform with our first agency customer, we found that customer-facing staff could not identify who had last worked an account roughly 70% of the time — a gap nobody in the business had named, in a market where a single point of retention is worth six figures.

That finding came from sitting with one frontline user for an hour. It reordered the product build and is now helping shape VelociBind →

TaskFreq.Owner
Renewal follow-upDailyRotates
Account handoff notePer touchUndocumented
Last touched by: unknown — ~70% of accounts

“I genuinely can’t tell you who spoke to this customer last.” — frontline staff, discovery interview

QuantizedIQ product

VelociBind

VelociBind is an insurance operations platform being designed from the outset to serve several independent agencies. Live operating workflows shape the product rather than assumptions about how agencies should work.

Beyond insurance

Other product directions

We are also currently in discussions around building products for the energy utility space and multi-user franchise operations. The common thread is the same: recurring operational problems that can become durable, reusable systems.

Morris Kabuage, founder of QuantizedIQ
Morris Kabuage — Founder, QuantizedIQ
About

QuantizedIQ builds automation systems and software products around the way work actually happens.

It was founded by Morris Kabuage, most recently a Senior Director in Microsoft’s Office of Applied Research — the group that moves AI from research into products people actually use. The work spanned Office, Windows, Teams, and SharePoint, and mostly came down to the unglamorous part: understanding how people really used a product before deciding what the model should do.

Across more than twenty-five years in technology, Morris has seen that gap from multiple sides: building SQL Server as a Microsoft program manager; watching how the business bought and used technology as a field business analyst; producing training data for large AI systems at Mighty AI, later acquired by Uber; and helping move applied AI into Office, Windows, Teams, and SharePoint.

Those vantage points shape how QuantizedIQ builds today: observe the work, identify the friction worth fixing, build against a measurable baseline, and turn repeatable problems into durable systems and products.

Watch the work first. Build second.

Based in Newcastle, Washington, working across the Puget Sound region and remotely.

Program Manager
SQL Server, Microsoft
Building the product.
Business Analyst
MCAPS, Microsoft
Watching how the business actually bought and used it.
Solutions Architect
Mighty AI (acquired by Uber)
Producing training data behind some of the world’s largest AI models.
Senior Director
Office of Applied Research, Microsoft
Getting AI into Office, Windows, Teams, and SharePoint.
Founder
QuantizedIQ
Turning observed workflows into automation systems and products.
Get started

Start with the map.

Bring one process that is costing your team time, money, or patience. We’ll talk through how it works today and whether an Operations Diagnostic is worth doing.

One process. No retainer and no commitment past the deliverable.