AI Transformation + AI Engineering
ACG helps mid-market companies set the strategy, change how teams work, and build AI systems that do real work.
What we do
AI Transformation →
Strategy, roadmap, and adoption with ACG at the leadership table.
AI Engineering →
Agents, pipelines, and platforms built into production.
Quick wins in the first 30 days.
You own everything we build.
Senior operators. No junior bench.
Two ways we help
Strategy + adoption
We help leadership decide where AI matters, build a roadmap the business can act on, and change how work gets done. Start with a workshop or assessment, then keep us at the leadership table through an embedded monthly partnership.
Systems + engineers
We build agents, document pipelines, internal tools, and custom platforms around real workflows. Choose a scoped project build or place a senior AI engineer inside your team on a monthly basis.
The stakes
Models are capable. Tools are everywhere. The hard part is choosing the right work, getting people to change their habits, and putting reliable systems into production.
A pilot that never changes a workflow is still a pilot. A license nobody uses is still overhead. The companies pulling ahead are connecting leadership decisions to deployed systems.
Selected companies we've worked with
Client voices
“Within 60 days, we went from zero AI strategy to three pilots running in production. The speed and clarity they brought was unlike any consulting engagement we've experienced.”
“They didn't just give us a slide deck. They sat with our teams and built the solutions alongside us. That hands-on approach made all the difference.”
“The leadership facilitation session completely changed how our executive team thinks about AI. We went from fear and skepticism to a shared vision in a single day.”
Results
Strategy earns its keep when the plan gets approved, the system ships, or the way people work changes.
$24M roadmap, board-approved
We interviewed stakeholders across the organization, assessed AI readiness, built the governance model, and designed a multi-year roadmap. The board approved the resulting $24M transformation plan.
100+ opportunities, agents deploying quarterly
We embedded with a 310-store retailer, interviewed every department, and scored more than 100 AI opportunities. The work now continues through an ongoing transformation partnership and embedded AI engineering.
Portfolio-wide AI training & board-level strategy
We built AI literacy across a private equity firm and its portfolio through executive training, board presentations, and an HR summit. The work became a repeatable model for portfolio-company AI readiness.
78% to 95%+ accuracy, 40% cost reduction
We built a document pipeline that routes different work to the model best suited for it. Extraction accuracy rose from 78% to more than 95%, while processing costs dropped 40%.
Governance, legal AI pilot, cloud migration support
We joined the leadership team as its embedded AI partner. The work covered stakeholder interviews, an AI governance policy, a legal AI pilot, and support for a cloud document-management migration.
Multi-tenant SaaS with AI-powered parsing & screening
We built a multi-tenant recruiting platform with AI-powered job parsing, candidate screening, and pipeline management. It is a working product, not a prototype or a slide deck.
The Signal
Robbie's notes on AI, adoption, and what changes when companies move beyond experiments.
Sports has leapfrogged corporate analytics despite having fewer resources—proving that innovation accelerates with tight feedback loops and obsessive practitioners. As AI democratizes sophisticated analysis, businesses finally have the tools to reclaim their competitive edge in data-driven decision making.
Read on LinkedIn →
GitHub commit graphs are becoming a public record of who's adopted AI—and hiring managers are starting to read them that way. This shift from invisible to measurable adoption has real implications for how companies identify and evaluate talent embracing AI tools.
Read on LinkedIn →
Forget the AI vision—pick one broken process with a clear cost and fix it in 4-8 weeks. One ACG client reduced invoice processing time by 70% in six weeks by focusing on a single high-impact problem, building internal confidence for the next project.
Read on LinkedIn →
An executive called Claude her "new boyfriend" after it solved a years-long document management problem in minutes. This consultant's insight flips the AI adoption playbook: skip the strategy decks and find the unglamorous tasks that actually convert skeptics into believers.
Read on LinkedIn →
It's rarely about skills or fear—it's ego. This consultant shares how peer-to-peer AI training, not mandates or workshops, actually drives adoption by reframing the technology as a bandwidth multiplier, not a threat to expertise.
Read on LinkedIn →
Organizations initially dismiss the need for AI engineers, then experience a dramatic shift in priorities within months of seeing results. The real challenge isn't selling companies on AI's value—it's helping them build strategically instead of rushing into overexpansion.
Read on LinkedIn →
Not every AI problem needs a closed feedback loop. When metrics are fuzzy, the smarter play is generating multiple variations and applying human judgment—like a photographer taking dozens of shots instead of one perfect frame.
Read on LinkedIn →
Seasoned experts aren't obsolete in the AI era—they're more valuable than ever. The ability to recognize when AI outputs are wrong, guide corrections, and know what good looks like is exactly what a decade of domain expertise provides.
Read on LinkedIn →
Custom software shops are eliminating build fees entirely, collapsing traditional project pricing into ongoing maintenance agreements. As AI makes initial development cheaper, the real cost—and real value—now lies in managing non-deterministic systems over time.
Read on LinkedIn →
Field Notes
Weekly notes on what is working, what is getting stuck, and what we are learning while helping companies put AI into practice.
“Every complicated legacy workflow feels like a requirement until you ask whether anyone would design it that way from scratch.”
Reviewed an integration with a client this week where the sticking point wasn't the technology, it was the approvals. They had a legacy system with sign-off rules layered by location, by role, and by individual person, with rules about which rule wins when they conflict. The team treated all of it as a hard requirement because it was how they had always worked. We can build almost any of it, but building the per-person exceptions on top of the per-office logic makes the whole thing brittle. What I kept wanting to ask was whether they actually needed it, or whether they were just carrying it forward. Every complicated legacy workflow feels like a requirement until you ask whether anyone would design it that way from scratch. Migrating a process is a good moment to question it.
“A client who had never written a line of code suddenly needed opinions on repositories, key rotation, and enterprise licenses, and none of that was on their radar a year ago.”
Spent time with a client this week working through the plumbing of building software, something they never set out to do. We were sorting through version control, how to store API keys safely, whether to buy the basic enterprise license or the fancier one. A client who had never written a line of code suddenly needed opinions on repositories, key rotation, and enterprise licenses, and none of that was on their radar a year ago. What struck me is how fast an organization crosses from we do not build software to we have a development practice with governance questions. AI lowered the cost of starting, but it did not remove the operational weight that follows. Once you ship something real, you inherit all the unglamorous parts. The build got cheaper. The responsibility did not.
“The deal used to be you pay the expert half the upside, now it's an AI tool plus the one person inside who actually knows your business.”
Sat in on an assessment review where a client asked who they should hire for a narrow optimization problem. The honest answer surprised them. I told them about a recent project where a company had once paid a domain specialist a large cut of the savings he negotiated for them. This time around, the work got done by me plus the right tools, paired with their internal person who understood how things actually worked. The deal used to be you pay the expert half the upside, now it's an AI tool plus the one person inside who actually knows your business. The expensive part was never the knowledge. It was the access and the time to apply it. Both of those just got a lot cheaper.
“It's a bit of a Pandora's box, there's no closing it now, and the question becomes whether you'd refuse to flip on a light switch.”
Reviewed a rollout with a client this week where someone on the team declined to participate in testing a new tool, citing ethical and environmental concerns. The manager wanted to know how to handle it without steamrolling the person. I don't have a clean answer, and I said so. It's a bit of a Pandora's box, there's no closing it now, and the question becomes whether you'd refuse to flip on a light switch. The comparison I keep coming back to is electricity arriving a century ago. You could opt out, but it was everywhere, and opting out mostly meant sitting in the dark. The honest move is to actually hear the concern first, then explain how little the tool changes their day, before deciding anything.
“If we were building this on Claude, we'd be on version three by now. With Copilot Studio, we're still trying to get the thing to show up in the app.”
Spent part of this week helping a client get a custom Copilot agent working inside Microsoft Teams. Should have been simple. It wasn't. The agent showed up in the web version but not the desktop app. Permissions kept resetting. The publish flow sent us in circles. Admin couldn't explain why the enterprise license was blocking it. The whole session was a live demonstration of what happens when a platform has every resource in the world and still can't ship a smooth developer experience. What struck me wasn't the bugs -- it's the contrast. When I work with Claude-based tooling, iteration is fast. You hit a wall, you work around it, you ship. With the Microsoft stack, every step has a different UI, a different permission model, and a different team that owns it. And yet most enterprise AI budgets are going to Microsoft, because that's where the existing agreements are. I understand the logic. I don't agree with it. The best AI tool isn't the one already in your contract. It's the one your team can actually use. Companies that figure that out early are going to move faster than the ones still waiting on Copilot to catch up to what Claude could do today.
“If my productivity can go from 1x to 10x, so can the bad guys' productivity.”
Preparing a board presentation this week and working through how to frame AI risk in a way that resonates with non-technical directors. One board member was especially focused on bad actors and what the organization should worry about. I've landed on a framing that clicks with people. If my productivity can go from 1x to 10x, so can the bad guys' productivity. But it's not just that existing threats get faster. The bar to become a bad actor drops dramatically. I ran a simple network scan on my home setup using an AI coding tool. It came back 30 minutes later with everything mapped out. No specialized training needed. That example tends to land harder than abstract warnings about cybersecurity. When a board member can picture someone with no technical background doing real damage, the risk stops feeling theoretical.
Why ACG
After 20+ years building AI companies and 170+ startup investments, we kept seeing the same gap: companies know AI matters, but they cannot get it to work inside the business. That's the gap ACG fills.
3 AI companies founded and acquired
20+ years building AI systems before the current hype cycle
Senior operators only. No junior consultant bench.
Everything is built in your stack, under your accounts.
Founder
Managing Director
Founded Automated Insights, the first generative AI company, before co-founding Infinia ML and Bionic Health. Two Masters from MIT. Cisco's youngest Distinguished Engineer. 8 AI patents. 170+ startup investments.
Executive Director
Serial entrepreneur. Founded and scaled ChannelAdvisor from startup to IPO (NYSE: ECOM). Brings deep experience in building and scaling technology companies.
LinkedIn →
Executive Director
Operations and growth leader. Brings deep experience in scaling professional services firms and managing complex client engagements.
LinkedIn →
Executive Director
Leadership and organizational change expert. Helps executive teams navigate the cultural and operational shifts that AI adoption requires.
LinkedIn →
Director of AI Transformation
An AI product leader, systems architect, and hands-on operator who turns complex workflows into production-ready intelligent systems. Builds automations, agentic workflows, and internal tools across SaaS, data, and operations to move teams from AI experimentation to measurable business outcomes.
Executive Director
Revenue and marketing strategist. Builds go-to-market engines and helps clients think about AI's impact on their customer-facing operations.
LinkedIn →
Executive Director
AI and enterprise technology specialist. Designs and deploys the agent systems, integrations, and technical infrastructure for client engagements.
LinkedIn →
Research Associate
Researcher and writer with a background in philosophy, computer science, and legal research. Supports client engagements with analysis, writing, and project research.
LinkedIn →FAQ
The next move
Bring us the business problem. We will help you decide whether the next move is a workshop, an assessment, a build, or an ongoing partnership.
Book a Call
