~20
active agents
Across LFG and its sister company, up from a single pilot agent in the spring.
Organizational Intelligence
From One Pilot Agent to an AI-Capable Organization
Organizational Intelligence in practice
How Landfill Group moved from zero production AI to employee-driven adoption across multiple business functions.
- Starting point
- A single pilot agent and no AI tooling in production across the business.
- Time period
- Under six months, from spring through midsummer 2026.
- What changed
- Roughly 20 agents across finance, legal, operations, sales, admin, and the fab shop — plus a sister company.
- Measured evidence
- The client assumed its own usage billing, demand exceeded the retainer two months running, and reporting output was rated better than the prior manual process.
- Still being measured
- Per-worker time savings are not yet instrumented. A usage-analytics workstream is on the roadmap.
- Adoption pattern
- Employee-driven: staff build their own workflows and ask to join the AI Council.
01—The headline
From zero AI tooling to employee-driven adoption in under six months.
Landfill Group is a successful operating business, not a technology company. In the spring it had a single pilot agent and no AI tooling in production. Within six months it was running roughly twenty agents across finance, legal, operations, sales, admin, and its fab shop — plus its sister company.
The number that matters more than the agent count is who is driving it. Staff are building their own workflows, asking to join the AI Council, and the company took over paying its own usage costs.
Adoption stopped being something we delivered and became something they pulled.
12+
employees with their own agents
Spanning finance, accounting, operations, legal, sales, admin, and the fab shop.
<6
months from zero
From no AI tooling in the business to org-wide, employee-driven adoption.
~1 hr
to activate a new agent
Roughly $100–200 all-in, making the marginal cost of adding a worker trivial.
02—Commitment signals
The measures that are harder to manufacture
Usage counts can be inflated by enthusiasm. These are the signals that cost the client something, which is what makes them worth reporting.
The client took over its own usage billing
Usage has roughly doubled month over month since.
Employees are asking to join
Staff have proactively asked for seats after seeing colleagues' results. Adoption is being pulled through the organization, not pushed onto it.
Staff are building their own workflows
Rather than waiting for handed-down automations, employees are designing their own agent workflows and documenting them so the work can be handed over.
03—Three examples
Where the work actually landed
SIMPLIFY
Finance reporting hands itself over
Monthly budget and AP reporting, rebuilt end to end by the person who owned it.
LFG's AP/Finance lead rebuilt monthly budget reporting on her own agent: raw QuickBooks exports go in, and formatted budget reports with favorable and unfavorable highlights come out, filed automatically to SharePoint.
What makes this a capability rather than an automation is how it happened. She designed the workflow herself, then asked Claude to document how she would teach it to someone else — and handed the whole process to the agent.
Asked whether the agent did the job as well as she had done it herself, her assessment was that it was technically better.
— AP/Finance lead, Landfill Group
SEE
Part numbering on the shop floor
A data cleanup that turned into the engagement's largest production workload.
The fab shop's Job Boss data cleanup became an AI-accelerated project: agents were used to design and migrate a new part-numbering system, with more than 36 hours of data migration executed with agent assistance in a single month.
The resulting agent now handles vendor pricing uploads, creates part numbers according to the naming convention, and maintains pricing archives. Scheduled supplier outreach is the next step.
Roughly 90% of early AI usage spend went to this effort — an indicator of real production work rather than experimentation.
SCALE
Absorbing a departing role
The clearest ROI storyline in the engagement.
When a departing employee's responsibilities needed a home, LFG chose to absorb the role with a combination of Mountain Dev and agents rather than rehire.
Agent-assisted takeover meetings and workflow documentation are underway, with the expectation of being net cash-positive against the salary it replaces.
AI adoption directly offsetting headcount cost, rather than adding to it.
04—Incremental benefit
What the business got out of it
Efficiency
Monthly budget and AP reporting automated end to end, with output quality rated better than the prior manual process.
Cost avoidance
A departing employee's role absorbed by agents and Mountain Dev, at expected net-positive cash versus the salary.
Sales leverage
75-page RFPs scored against a matrix, with responses drafted to roughly 80% complete.
Risk visibility
Regulatory and permit data, including state notices of violation, pulled into dashboards for project evaluation.
Time recovered
The integration agent alone unblocks about two hours a week; email triage and spreadsheet updates in community impact are now agent-handled.
Reach
Adoption extended beyond LFG into its sister company, covering retail and hospitality staff across three locations.
Per-worker time savings are not yet instrumented across the program. A usage-analytics workstream is on the roadmap, and we would rather report the measures we can stand behind than estimate the ones we cannot.
05—Our own journey
We are running this on ourselves first.
Mountain Dev has spent years building software for other organizations. Organizational Intelligence is the first practice we have built by applying the framework to our own business before selling it — mapping how our own work flows, finding where our own process was the constraint, and instrumenting what we could measure.
That is also why this page is shorter than most case study pages. The engagement is real and ongoing, the numbers above are the ones we can substantiate today, and we would rather publish a narrow set of defensible results than a broad set of impressive ones.
The evidence should outlast the enthusiasm.
Where to go next
See whether the same approach fits your business
Start with clarity
See how the work is structured: a low-risk assessment that maps workflows and prioritizes opportunities, then quick wins that prove value before anything larger is committed to.
See how we workStart with a conversation
A focused executive discussion about your goals, current systems, and whether there is a sensible low-risk place to begin.
Start a ConversationMountain Dev
Building Organizational Intelligence.