A Black Belt Approach
to AI That Gets Used
Turning ARCO/Murray’s AI investment into real adoption across 70 salespeople and 50 offices, using the same disciplined methodology your engineers already trust.
Before you read, a few minutes on how we’d approach this
A short note on who we are, the principles behind the work, and how we’d build this with your team.
You do not have a technology problem
ARCO/Murray already has the pieces most firms are still shopping for. You have engineers who can build. You have a platform. What decides whether any of it pays off is adoption, whether 70 proud, independent salespeople across 50 offices actually change how they work. That is the hard part. It is the part your team told us they need help with. It is the part we do.
You have built a real internal capability in ARCO Murray Innovation, committed to Glean, and put dedicated people against sales enablement. That is the right foundation, and we are not here to rebuild it. The gap is not the technology.
That sentence is the entire reason for this proposal. Getting a decentralized sales force to adopt new tools is not a software task. It is a change-management project repeated across 50 offices, each with its own P&L, its own leaders, and salespeople who range from 24-year-old rookies to 30-year veterans who have earned the right to be skeptical. We run that project the way your firm runs everything else: as a disciplined, measured engagement with a beginning, a proof, and a controlled rollout.
A Six Sigma Black Belt methodology for AI adoption, built on the DMAIC structure your engineers already respect. A senior team that has moved distributed workforces through exactly this kind of change. And the translation layer between your technical group and your sales floor that turns what the field needs into what gets built, and back again.
We propose a fixed $50,000 engagement, total for two months, that ends with one agent live with your power users and hard adoption numbers. From there, a $20,000 per month retainer runs the rollout, on an initial six-month term. That is a fraction of what the national firms quoted you, and unlike a slide deck it ends with something your people are actually using.
The Problem
What we heard from you, and what we believe it actually means. If any of this is wrong, it changes the plan, and we would rather know now than be polite about it.
What we heard
Before we recommend anything, we want to prove we listened. Here is your situation as we understand it.
The firm
ARCO/Murray will do roughly $7 billion in revenue this year across 50 offices in the US and Canada, with more than a thousand employees. You are deliberately decentralized: every office runs its own P&L, with its own leaders, salespeople, and project managers, and its own way of doing business. You are design-builders who only hire engineers, get involved early to design a better mousetrap, take on risk others will not, and win a large share of repeat work because of your national reach. You bring a premium capability but compete on value rather than sticker price, which means your edge is getting in early and engineering a smarter, more cost-effective build.
The sales organization
Seventy salespeople, one or two per office on average, more at the Chicago headquarters, ranging in age from 24 to 60 and in experience from first-year to three decades. On a one-to-ten scale of self-reported AI fluency, a baseline we would set at kickoff, most of your people sit between a three and a seven. The goal, set from the top, is to pull the threes and fours up to fives, and the fives up to sevens, and to accept that a few will be nines and a few will stay fives. Leadership is committed and clear-eyed: if ARCO is not leading on this, a competitor is, and they will use it to take your lunch.
Where your AI effort stands today
You have stood up ARCO Murray Innovation under a strong technical lead who can build agents and tracks where the technology is heading, and you have committed to Glean, a serious enterprise platform that already gives you a knowledge graph across your systems, an agent builder, model routing, and delivery through Teams. Around ten people have Claude access, others are on ChatGPT, and usage is spread out. Monthly AI-for-BD trainings have just started, and you have hired a talented young analyst to help drive sales adoption. You also looked hard at a full engagement with the national consulting and model firms and chose to pass, not for lack of budget, but because the price did not clearly map to the return.
You have the builders and the platform. What you do not yet have is a disciplined system for adoption across a decentralized firm, and a translator between your technical team and your sales floor. Asking a single analyst to carry a firm-wide, 50-office behavior change is a heavy load, and you already sense it. That is precisely the load we are built to share.
What we believe
Three convictions shape everything that follows. They are also the reasons the last mile, the part after the technology works, is where the money is won or lost.
1. The bottleneck is adoption, not capability.
Your innovation team can build almost anything you point them at. The open question is whether a top producer in one of your offices actually changes his Monday morning because of it. Across 50 offices with their own P&Ls, that is not one change-management problem. It is fifty. A corporate mandate alone does not fix it, because each office answers to its own margin. What moves them is peer proof: early wins in the first offices that make the next leader want in rather than comply. No agent, however clever, does that on its own.
2. AI only gets used when it meets people where they already are.
A new tool to log into, learn, and remember is a tax most salespeople will not pay, especially the veterans. The pattern that works is an assistant reachable through the tools your people already live in, their email, their Teams, a text message, that answers inside their context. No new habit. No new tab. That single design choice does more for adoption than any amount of training.
3. Value leaks out in the gap between your engineers and your sales floor.
Your technical lead thinks in architecture. Your reps think in relationships and deadlines. Somebody has to translate what the field actually needs into what gets built, and translate what is now possible back into language a skeptical thirty-year veteran will try. That translation layer is the difference between a tool that ships and a tool that sticks.
We are not here to replace ARCO Murray Innovation or to compete with Glean. We make both pay off. Glean is the platform. Your team builds. We bring the sales-domain expertise to design the agents worth building, and the adoption muscle to make the field actually use them. Most firms use a fraction of what a platform like Glean can do, because a platform is not a strategy and an agent builder is not a built agent.
The Method
You are an engineering firm. You only hire engineers. So we run this the way you would run any process improvement, with rigor and measurement rather than a pile of ideas and a hope that people use them.
The Black Belt methodology
We run every AI adoption engagement as a Six Sigma Black Belt project on the DMAIC structure, because the discipline that de-risks a process on the plant floor is what de-risks a technology rollout across 50 offices.
Agree on what winning looks like before anyone builds.
We align your COO, your sales leaders, and your innovation lead on the exact use cases, the teams in scope, and the metrics that matter: adoption-score movement, proposal turnaround time, and hours handed back to selling. We map the risks too, data sensitivity, change resistance, and model dependency, so there are no surprises later.
Baseline the truth, by office.
We quantify current token spend and tool sprawl across Claude, ChatGPT, and Glean, find where knowledge is fragmented, measure how much time reps lose re-explaining context to a chatbot, and capture the honest three-to-seven adoption distribution across offices. You cannot improve what you have not measured, and this baseline is what proves the ROI later.
Design the agents and the economics on top of what you already own.
We design the specific, high-value sales agents your business actually needs, the part a horizontal platform cannot do for you, and architect them to run inside Glean and reach reps through the tools they already use. We also tune model routing and caching within your platform so mission-critical work hits the best model and routine work hits a cheaper one, taking real cost off your token spend without adding another system to govern.
Launch the first agent with your power users, not a random office.
We stand up the first agent, designed by Icon and built by your innovation team in Glean, and launch it with the salespeople already at a seven or above, across offices rather than by geography. They adopt fastest, produce the first wins, and become the credible proof that pulls the skeptical majority along.
Make it stick across the firm.
This is the phase most efforts skip and most failures trace back to. Budget guardrails and model controls, role-based access, weekly optimization, and the human system that keeps adoption alive: the change-management and training engine that a single analyst cannot carry alone.
What we help you build
The sales-facing agents we would design, prioritize, and drive adoption of, working alongside your innovation team and Glean. Not a shopping list. A sequenced set that puts wins in front of your reps first.
Watches permits, land acquisitions, REIT and private equity purchases, hyperscaler and data center announcements, and corporate expansion news across every ARCO market, so your reps get into a pursuit before the formal request ever exists.
Scores each opportunity against ARCO’s real win profile: sector fit, project size, margin pattern, and whether it touches an existing relationship. Tells a rep where the week is best spent.
Builds a live dossier on every target and account, decision-makers, prior facilities, who they have built with before, and the warmest path in through ARCO’s own network, so every rep walks into every meeting briefed.
Drafts first-pass qualifications and design-build proposals in hours by pulling the most relevant past projects and differentiators, so your people spend their time selling instead of assembling decks.
Continuously watches existing clients for expansion signals, new site searches, renewals, portfolio moves, and nudges the right rep at the right moment, aimed straight at the repeat work your national reach already wins.
Identifies satisfied clients at project close, routes reference and referral requests, and keeps ARCO top of mind with the developers and brokers who feed you work.
We design and direct the agents; your team builds them.
Icon and our technical lead design each agent for the construction-sales world and architect how it should work, then direct the build shoulder to shoulder with your innovation team, who build it inside Glean. You get domain thinking and senior direction from us, and your team keeps ownership, control, and the credit. No shadow systems, no lock-in, nothing new to govern.
We drive adoption where the platform stops.
Glean can measure adoption. It cannot walk into 50 offices and change how a thirty-year veteran works. That is people work: role-specific rollout, sales-floor translation, and a training rhythm your team cannot carry alone. It is the part every AI program underinvests in, and the part that decides whether any of this shows up in the numbers.
Why this works where training alone will not
Training teaches a tool. Adoption requires the tool to fit the day. Those are different problems, and the second one is harder.
You already saw the limits firsthand. A training run by a couple of enthusiastic power users tends to lose the room, because the way a ten-out-of-ten uses AI is not where a four can start. And putting a firm-wide behavior change on the shoulders of a single analyst, however capable, is a lot to ask. None of that is a criticism of your people. It is the nature of change across a proud, decentralized sales force.
We sequence it the way adoption actually works. We start with your power users, the people already at a seven or above, across offices rather than by geography. They adopt fastest and generate the first real wins, and a respected veteran saying this helped me win moves the skeptical majority in a way no outside firm ever could. We use their credibility, but we own the program, so it does not become a couple of enthusiasts going on tangents and losing the room.
Think of us as the general contractor for adoption. Your innovation team supplies the engineering. Glean supplies the materials. We sequence the work, translate between the trades and the field, and stay on the job until it is actually being used across all 50 offices. That is the difference between an AI program that photographs well and one that shows up in the numbers.
How we measure it
Every metric below is baselined in the Define phase, so movement is measured against your real starting point rather than a number we invented. These are the numbers we report on, and the numbers we expect to be held to.
How Icon plugs into your team, not over it
Three parties, one system. Glean is the platform you have chosen. Your innovation team builds and runs. Icon brings the construction-sales expertise to design the agents worth building and the muscle to make 50 offices actually use them.
- Designs the agents your sales business actually needs, from a construction-sales operator’s point of view
- Architects and directs the build, shoulder to shoulder with your team
- Translates between your technical team and your sales floor
- Drives adoption, power users first, across offices
- Runs DMAIC governance and measures what is working
- Builds each agent for scale inside Glean
- Connects data and systems across your platform
- Secures, scales, and maintains the agents in production
- Owns the platform relationship and the technical standard
- Keeps control and the credit for what ships to the field
- Knowledge graph across your systems, the company brain
- Agent runtime and builder for production deployment
- Model gateway for routing and cost control
- Delivery through Teams so agents reach reps where they work
- Adoption and governance metrics at the platform level
Icon designsfrom the sales floor’s needs
Icon directsarchitects how it works
Team buildsfor scale in Glean
Deploy to fieldthrough Teams, no new tab
Icon drives adoptionpower users first
Measure & iteratenext agent enters the loop
The Team & The Offer
Who does the work, what they have already done, and what it costs. Two numbers, both fixed, with the hardest part included rather than sold separately.
The team on your engagement
You are not hiring a tool and a login. You are getting a senior team of three, embedded as your fractional AI implementation group, for exactly as long as you need us and without the cost of three full-time specialists who would run north of a million dollars a year.
The hard part of this work is not knowing construction. You have more than a thousand people who know design-build better than any outside firm ever will. The hard part is translating what the sales floor actually needs into agents that get used, and staying on the job until the numbers move. That is what these four have spent their careers on.
Daniel Askew
A multi-business founder who builds, buys, and operates companies across real estate, financial services, technology, and hospitality, and advises dozens more. He got obsessed with AI early and built Icon’s own multi-agent operating system from scratch, the company brain this practice grew out of. Your day-to-day partner and the person in the room for every decision that matters. Owns the what, the why, and the relationship.
Rachel Askew
Rachel is Daniel’s business partner and provides senior oversight from behind the scenes. She is not a billed member of the engagement team, but she stays close to the work because it matters to us that it lands. She has led change management and operational transformation inside billion-dollar enterprises moving through PE transactions and mergers, the kind of work where thousands of people have to change how they operate at once.
Sandeep Sardana
Twenty-five years leading large-scale technology and transformation programs across financial services, healthcare, and life sciences. Delivered six times faster time to market and more than $50 million in savings at a top-five pharmaceutical company, and cut deployment lead times by more than 30 percent in a large insurance transformation. Works peer to peer with your innovation lead and owns the architecture and the discipline that makes adoption stick.
Joel Campbell
Career built at the intersection of technology and the people who have to use it. Built the internal operating platform that scaled a healthcare services company from zero to tens of thousands of member interactions per month, led clinical system rollouts, and shipped AI workflows that improved productivity and staff satisfaction at the same time. Owns the requirements, the adoption program, and the translation between what the technology can do and what a rep will actually use.
Across the four of us: a firm-wide AI operating system built and run in production, more than $50 million saved on a top-five pharma program, large-scale change delivered inside distributed financial-services and life-sciences organizations, and a career spent building the products sales and operations teams actually adopt.
Investment
Two numbers, both fixed and known before work begins. No hourly meters, no surprises, and the hardest part of the work is included rather than sold as an upgrade.
| Engagement | What You Get | Timeline | Investment |
|---|---|---|---|
| Blueprint & Launch | Five deliverables: the success metrics agreed with your leaders, a measured baseline of current spend and adoption, a prioritized roadmap for the sales organization, the integration plan with your innovation team, and one agent designed by Icon, built with your team in Glean, and launched live with your power users. | Two-month engagement | $50,000fixed, total |
| Control Retainer | Your Icon team embedded to run the rollout across the sales team: a defined cadence of new agents each quarter, adoption cadence, governance, and cost optimization. Initial four-month term, with a six-month renewal option. Scope is the sales department, with an option to expand to operations or other functions as the work proves out. | Ongoing, after launch | $20,000/ mo · initial 4-mo term |
In plain numbers: $50,000 total for the two-month launch, then $20,000 a month. The initial retainer term is four months, with a six-month renewal option. You are all in at $130,000 through the first six months. Fully renewed, the complete twelve-month engagement runs $250,000.
The national firms scope this kind of work into the high six and seven figures. We would rather earn the relationship by proving the return with your first users than sell you a large number on faith. If the launch does not show real adoption, you are under no obligation to continue. The retainer exists only for as long as it keeps earning its place.
We price to win your business, not to hedge, so nothing here puts our fee at risk. But we are confident enough in the work that at the end of the launch, and again at the end of year one, we invite a discretionary success bonus in an amount you determine, if you believe we have earned it. Measurable adoption movement across offices, pursuits your reps would otherwise have missed, and real cost taken off your token spend are the outcomes that would merit it. The decision, and the number, are entirely yours.
The return, framed conservatively
We will not hand you a fabricated multiple. On a $7 billion revenue base, this engagement does not need to transform the firm to pay for itself many times over. It needs to help a meaningful share of your 70 reps get into pursuits a little earlier, respond a little faster, and spend more of their week in front of clients instead of assembling decks. Move adoption across even a few offices from a four to a six, and the return dwarfs a $250,000 engagement. Everything past that compounds.
Third-party software and infrastructure that ARCO owns are billed at cost with full transparency and never marked up. Final scope is confirmed in the Define phase.
What happens next
This document is our read after one conversation. The next step is the one that makes it yours.
A working session with your team
Bring us back in with Jason, your innovation lead, and the rest of the AI leadership team. Everything here is built on a single conversation with Morgan, which means we are making informed assumptions about your business. That session is where we replace them with what your people actually know. Book the working session.
Green-light the Blueprint
We start with Define and Measure inside the first week, aligning on metrics and baselining current spend, sprawl, and adoption by office.
Watch it work
Two months later you are looking at one agent live with your power users, real adoption numbers, and a company-wide plan for scaling across all 50 offices.
Prepared by Icon AI, a service line of Icon Business Advisors, Nashville, Tennessee. Confidential and intended solely for ARCO/Murray National Construction Company. The thinking, frameworks, and methodology in this document remain the property of Icon Business Advisors. Client engagements referenced were performed by individual team members during their prior consulting careers. Cost figures reflect public 2026 provider pricing and industry reporting and are cited as directional context. All investment figures are fixed as stated.