Warm furniture showroom at golden hour with staged living and dining vignettes

Prepared for Tepperman’s

You have stronger AI foundations
than most companies at this stage,
and less coordination holding them together.

An AI Readiness Assessment prepared by Scott Walker, UpShift Collective

July 2026 · Confidential

0: Years in business
0: People using Claude today
0: Readiness areas assessed
0: Opportunities identified

01 · Purpose

An assessment, not a proposal.

A shared picture of where Tepperman’s stands on AI today, and enough structure to decide what to take on first. Everything here traces to something your team said in one of two discovery sessions, or to the technology inventory Adam provided. Where we could not verify something, we say so.

100

Years in business

People using Claude today

Readiness areas assessed

02 · Themes

What We Heard

Four themes, consistent across both sessions. Each closes with what we take it to mean.

The adoption gap is a structure problem, not an interest problem.
24 people use Claude today; Masoud estimates ~100 could benefit. Skepticism has already flipped positive, and what remains is unfamiliarity rather than fear. The first SME cohort still stalled — as Adam put it, staff were handed a plot of land and told to build a house with no foundation to start from.
What this means: the raw material is there. The missing piece is structure that makes it compound.
Data quality sits underneath almost everything.
Custom fields may not be reflected in NSAW, and fifty-six scripts fire on one sales order, conflicting and generating bad data downstream. Many departmental KPIs still are not reported a year in, on tooling you already own.
What this means: an AI reporting layer built on untrustworthy data will produce confident, wrong answers — and cost credibility when you need it most.
The manual load is large, well understood, and unevenly spread.
Reconciliation is a daily hand-worked exception list Darryl called ripe for AI. Thousands of calls a day are all answered by a person, and he wants a 30% cut in triage volume now. Service, delivery and credit are automated only at the routing layer.
What this means: there is no shortage of candidates. The choice should turn on what can move now, not on what is largest.
The competitive gap is real, and your team named it first.
Ashley and The Brick are both running better AI chat and triage today, and Andrew's point is that other retailers are already cutting call volume. The website has been costing the business since migration, with audits surfacing problems and no sustained plan to act.
What this means: you do not need to imagine the value. Your leadership has already identified where you are behind.

A Tepperman's sales associate with a tablet helping a couple on the showroom floor

What we heard

The people are the advantage. AI only pays off if it makes their day easier.

A recurring theme across the leadership conversations

03 · Readiness

Where You Stand Today

Seven areas, described rather than scored. Sponsorship and tooling sit well ahead of governance, data and measurement — that gap is the diagnosis. Open any area for the evidence behind it.

What we heard
Five senior leaders — Andrew Tepperman, Masoud Negad, Darryl Burgess, Jerry Martel and Adam Hansen — attended the full July 3 session, and all participated.
Vision 2035 names technology and customer experience as guiding principles, so AI attaches to existing strategy rather than competing with it.
The themes were consistent regardless of who was speaking or which department they represented — including the competitive gap, which the team named unprompted.
What this means: you are not starting by building the business case. That work is done.
What we heard
24 people use Claude today, connected directly into NetSuite through a read-only MCP connection.
Three Microsoft Copilot licences ride along with the M365 E3 stack — limited, with no agent capability.
23 of 24 inventoried systems integrate with NetSuite, so the data path for most use cases already exists.
Frontline IQ, the AI sales roleplay platform provided through Ashley's supplier network, is live with management backing and rolling out company-wide.
Masoud estimates roughly 100 people could realistically benefit from access, which leaves about three-quarters of that group with no formal path in.
What this means: the capability is bought, connected and proven. The constraint is who is allowed in and how, not what you own.
What we heard
An in-house NetSuite developer works almost exclusively through Claude Code and Codex, using AI diagnostics to find root causes in scripting errors where one sales order triggers 56 separate scripts.
Fully AI-built dashboards are in use in departments including customer care and credit.
Wins are surfacing independently from people like Shannon in customer care and members of the credit team.
Supply chain requested licences after a single session; two new implementation partners are using AI themselves to diagnose ERP and web platform root causes.
What this means: you have internal proof points to point at. But the deepest capability sits with one person and is undocumented — a dependency, not a practice.
What we heard
24 of an estimated 100 knowledge workers have access, granted department by department through a champion-delegate model.
The first SME cohort was given Claude access and asked to play and report back every two weeks. It lost momentum. Adam's analogy: handed a plot of land and told to build a house, with no foundation to know where to start.
Masoud is clear the fix is not disbanding the SMEs — it is staying with the game plan and giving it the structure and accountability the first attempt lacked.
Adam has restarted bi-weekly departmental sessions, one in as of July 3. Verbal resistance has already turned positive; what remains is unfamiliarity, not fear.
What this means: the raw material is willing. What is missing is a foundation that makes early wins visible enough to sustain the habit.
What we heard
No written AI use policy exists — no acceptable-use standard, no rules on customer credit or bureau data, no tool selection criteria.
Access is granted informally, department by department, rather than through a coordinated rollout.
Darryl's read: most employees are already working outside the sanctioned environment.
What this means: this is the cheapest gap on the list to close, and closing it unlocks the access departments are already asking for.
What we heard
Darryl described more than 2,000 custom fields in NetSuite, many not reflected in NSAW, the analytics warehouse used for reporting.
Adam confirmed NSAW is essentially a synchronized copy of live NetSuite data — if a field is not already tracked, someone has to build the pipeline by hand.
56 scripts fire on a single sales order, conflicting and generating bad data downstream. In Adam's words from June 5, the tool has "suffered" as a result.
Two implementation partners were recently onboarded on six-week engagements — one on the SuiteCommerce Advanced web platform, one on ERP scripting.
What this means: AI built on this data will produce confident, wrong answers. Anything needing trustworthy reporting sits behind the remediation already underway — a reason to be selective, not to wait.
What we heard
Masoud was direct: a lot of the departmental KPIs needed to manage a good business simply are not reported today, a year into implementation.
Frontline teams already know they could be better operators if they could see the numbers in front of them.
No measured baseline exists for any process named in this assessment — not reconciliation hours, not call handling, not call review time.
What this means: without a baseline you cannot prove what AI changed. Establishing it is part of the first phase, not an afterthought.

04 · Map

The Opportunities

Thirteen, each raised by a named person in one of the two sessions. Open any row for what was said, the impact, and where starting would begin. The list is a map, not a to-do list — the useful part is what it helps you rule out.

A foundation under the restarted departmental sessions, plus named ownership, fixed cadence and accountability.
Why it came up
The first SME cohort was given Claude access and asked to play and report back every two weeks, and it lost momentum. Adam's analogy: staff were handed a plot of land and told to build a house, with no foundation to know where to start. Masoud is clear the fix is not disbanding the group — real wins are already surfacing on their own from Shannon in customer care and members of the credit team — it is structure and accountability.
Business impact
Turns unmanaged experimentation into a program where wins like supply chain's stay lit instead of fading, and where each department has one named owner with authority to change how work is done.
Starting point
Role-based curriculum behind Adam's existing bi-weekly cadence, with champions named per function and progress visible by department.
First-line AI triage on Contivio, against Darryl's 30% reduction target.
Why it came up
Andrew flagged the volume directly: thousands of calls arrive every day and every one is handled by a person. Darryl believes a basic triage layer on the existing Contivio contact centre platform could take out 30 percent of them without significant added complexity. Both he and Jerry named Ashley and The Brick as already running stronger AI chat and triage — a gap the team raised unprompted.
Business impact
A 30 percent reduction in triage-level calls returns hours to the care team every day and closes a competitive gap your own leadership identified first.
Starting point
Scope the top call reasons on Contivio, then pilot deflection on the two highest-volume, lowest-risk intents.
Transcription, summarization and scoring for tone and resolution, on Contivio's AI bundle you are already evaluating.
Why it came up
Call quality is assessed today by a person listening to calls, which means coverage is thin and feedback is inconsistent. The capability sits inside the Contivio AI bundle already under evaluation.
Business impact
Consistent scoring across every call rather than a sample, with specific coaching on tone and resolution quality — and no added QA headcount.
Starting point
Turn on transcription for one team, agree a scoring rubric, and review the first two weeks with supervisors.
Automating the run-compare-chase cycle so the team starts at the exception list, not raw data.
Why it came up
Darryl's team runs daily audits comparing multiple reports to shake out an exception list, then digs into each account by hand — in his words, "running your cursor over all the fields that are supposed to be there." He called it directly ripe for AI discussion, for both the analytical and the auditing power.
Business impact
Removes the daily report-compare-chase cycle and frees hours that have gone into reconciliation every week since the NetSuite go-live, moving the team to judgment work on exceptions only.
Starting point
Codify the comparison logic already run manually, then generate the exception list automatically with reasons attached.
Forecasting recoveries and cash flow across the loan book, plus decision support on non-auto-approved files.
Why it came up
Tepperman's is, in Darryl's words, the only retailer in the industry he knows of running its own full financing book — which is what makes offers like 60 months interest-free with no payments for 36 months possible. But adjudication of anything the third-party bureau does not auto-approve, and all portfolio forecasting and recovery analytics, happens entirely in-house and entirely by hand.
Business impact
Protects the genuine competitive differentiator by bringing forecasting and decision support to the manual core of it, with the same customer data also feeding marketing analytics.
Starting point
Model recovery and cash-flow forecasting on historical book performance before touching adjudication decisions.
Closing the reporting gap on Oracle Analytics Cloud and NSAW, which you own.
Why it came up
Masoud was direct: a lot of the departmental KPIs needed to manage a good business are simply not reported today, a year into implementation. Frontline teams already know they could be better operators if they could see the numbers. The obstacle is underneath — custom fields not reflected in NSAW.
Business impact
Gives departments the numbers they need to run their own areas, and creates the baseline any AI impact would later be measured against.
Starting point
Pick one department, name its five KPIs, and confirm whether the fields behind them exist in NSAW today.
Automating the manual layer around DispatchTrack: scheduling, time allotment, route checks, customer comms.
Why it came up
Jerry and Darryl separately described the same thing: technician visits are scheduled by hand and delivery trucks are routed by hand, with a person then sense-checking the plan before it goes out. Darryl called delivery and routing right for this area — one of the more contained, well-defined targets on the table.
Business impact
Automates scheduling and route optimization while keeping human review as the final check rather than removing it, on a process that is fully manual today.
Starting point
Automate first-pass routing and time allotment inside the existing DispatchTrack flow, human approval retained.

What we left out, and why

The technology inventory suggested three more — AXPER traffic against sales, Bazaarvoice reviews against vendor claims, AI exception review on Demand Solutions forecasts. Credible on paper, but nobody raised them, so they are named here rather than presented as findings.

05 · Sequencing

Where We Would Start

Thirteen is a map. The smallest set of moves that would change the trajectory is three: two use cases, and the people work that decides whether they still matter in a year.

Give the fluency and champions work real structure
The non-negotiable. Adam's bi-weekly sessions already exist; this is about a foundation under them plus named ownership, cadence and accountability. The first cohort failed because nobody owned whether it continued.
Put AI call quality review into customer care
The most contained option. It runs on Contivio, which you already license, addresses time cost and coaching inconsistency, and is internal-facing — so a poor result costs nothing with customers.
Pilot first-line call triage
Darryl wants this immediately at roughly 30% of triage calls. It closes a competitive gap your team named, and it is the first thing customers would notice — which is why it goes second, not first.

Why not the bigger items first

Reconciliation and credit portfolio work are larger, but both are entangled with the NetSuite remediation and slower to show a result than an organization coming off a stalled initiative can wait. Better second moves.

If your team weighs it differently

If internal time recovery matters more than the customer-facing gap, account audit replaces call triage as the second move — coordinated with your ERP partners so you are not automating a process about to change.

Before anything starts

Capture a baseline first. Nobody has told us how many hours reconciliation consumes or how long call review takes. The number you measure before you deploy is the only way to prove afterwards that it worked.

06 · Method

How This Would Work

The shape of the work, in four phases. Offered for context, not commitment.

01
01 · GOVERN
Week 1
Rules of the road
- Executive alignment session: what AI is for here, and what leaders will model
- Acceptable use, data handling, access tiers
02
02 · ENABLE
Weeks 2–6
Fluency in the people doing the work
- Master sessions — live, cohort-based, how to build a workflow
- Role-based curriculum and credentialing, progress visible by department
- Departmental application sessions on Adam's existing bi-weekly cadence
03
03 · DEPLOY
Weeks 4–8
Building the specific things
- One or two narrow use cases taken to production
- Measurement attached from the start
04
04 · HAND OFF
Ongoing
Owned by your team
- Champions named — one owner per function, with authority to change how work is done
- The goal of every engagement is to become unnecessary

Phases overlap rather than run end to end — enablement starts while governance is being finalized, and the first build starts before enablement finishes.

Where UpShift fits

Your team owns the decisions and the systems. We own the plan, the curriculum, and keeping it moving.

Own the plan

Turn thirteen opportunities into a sequenced roadmap with named owners, a measured baseline before anything is built, and a success measure per move.

Gap it answers: No written plan today, and a first SME cohort that stalled for lack of structure.

Coordinate with leadership

Run the working cadence with the IT steering committee, keep the executive team aligned on sequence and trade-offs, and surface blockers early rather than at review time.

Gap it answers: Five leaders engaged, no single accountable owner.

Lead the fluency curriculum

Design and deliver the relaunched program — executive sessions, role-based curriculum for managers, and departmental champions — with progress visible and accountability built in.

Gap it answers: 24 of roughly 100 potential users, access granted informally.

Set governance and guardrails

Draft the written policy with IT and finance: acceptable use, access tiers, and explicit rules for credit and financing data.

Gap it answers: No policy today; access runs on an informal champion-delegate model.

07 · Next

One working session. One decision.

A working session with your leadership team to go through the opportunities, the starting points and the questions, and land on one. The existing IT steering committee is the natural forum, with a single accountable owner coming out of it. That conversation is worth having whether or not UpShift is involved in what follows.

The choice worth making

One fork, and yours to pick — doing both at once is how organizations finish neither.

Card A

Build capability first

What it means here: AI fluency across the executive team and senior managers, with governance settled alongside it.
When it is the right call: When the constraint you most believe in is structural. Until people know what good looks like, use cases will keep being found by accident rather than on purpose.

Card B

Prove value first

What it means here: One or two specific use cases taken to production. Customer care is the obvious candidate given what Darryl said about triage volume and call quality review.
When it is the right call: When the constraint you most believe in is credibility. The organization has heard about AI and has not yet seen it do anything, and a visible result buys more goodwill than a training calendar.

Our view, offered as a view

A narrow version of both beats a full version of either. Settle governance — days of work, and it unlocks access departments are already asking for. Then pick two departments, not thirteen, and go deep enough to produce something people can see. Customer care and finance are the obvious two.

No pricing here, intentionally. Scope follows priority, and priority is yours to set.

Questions for your team

This assessment cannot answer these. How you do will set the starting point.

SEQUENCING
Should the first visible win be customer-facing or internal? Triage and call quality review imply different risk appetites.
DEPENDENCY
How much will you start before the ERP remediation lands? A conservative answer narrows the field considerably.
AUTHORITY
Can champions change how their department works, or only share what they learn? The first cohort had only the second.
EVIDENCE
What would you need to see in sixty days to justify going further? Name it before you start.

“What struck me is how much you already have going for you: real executive engagement, tools in place, a cleaner stack than you probably realize, and people finding wins unprompted. The gap is not capability. It is coordination — the easier of the two to solve.”

– Scott Walker, Founder & CEO, UpShift Collective

You have stronger AI foundations than most companies at this stage, and less coordination holding them together.

An assessment, not a proposal.

What We Heard

The adoption gap is a structure problem, not an interest problem.

Data quality sits underneath almost everything.

The manual load is large, well understood, and unevenly spread.

The competitive gap is real, and your team named it first.