01 The problem
An agency that automates other companies' operations has no excuse for running its own out of spreadsheets and browser tabs. For a while, we did anyway: revenue in one place, campaign state in another, workflow results in a third, and the true picture of the business assembled by hand whenever someone needed it.
That is the same disease we diagnose in clients, and it has the same cost: hours of assembly work around minutes of judgment, decisions made on stale numbers, and operational knowledge living in whoever last looked. So we treated ourselves as a client. This case is the result, and it matters to anyone evaluating us for one reason: the architecture we sell is the architecture we trust our own books to.
02 The architecture: directives, orchestration, execution
Everything in this build runs on the three-layer architecture we ship to clients, which we call DOE.
Directives are plain-text SOPs that define what should happen: how leads are scraped, how replies are handled, how proposals are generated. Orchestration is an AI agent that reads those directives, makes decisions, and routes work. Execution is deterministic Python, one script per job, doing the actual API calls and data processing.
The reasoning is arithmetic. An agent that is 90 percent reliable per step is 59 percent reliable across five chained steps. Push the steps into deterministic code and the agent only makes decisions, not mistakes of execution. The agent layer stays thin on purpose; the 50+ scripts below it are boring, testable, and correct, which is exactly what you want from the layer that touches money and outreach.
03 What we built
One dashboard at the center of the agency, with ten operating surfaces: a KPI overview, finances with P&L and invoices, team management, a client overview, the lead database, campaign tracking, workflow run history, workflow templates, job-sourcing pipeline, and settings.
Basis for the counts: routes and scripts in the production repository, June 2026.
Under the dashboard sits the execution layer, 50+ Python scripts in production covering lead generation, email enrichment, reply detection, AI strategy and deck generation, spreadsheet operations and self-healing utilities. The scripts run serverlessly on Modal behind a webhook layer, so a campaign event or a schedule can trigger work without anyone at a keyboard, and every run lands back in the dashboard as inspectable history with logs.
The financial surfaces read from the same database the automation writes to. Revenue, costs and client state are not assembled for meetings. They are simply current.
04 The technologies behind it
- Next.js for the dashboard, the same App Router architecture as our client platforms.
- Supabase as the operational database: users, clients, payments, expenses, leads, campaigns, runs.
- Modal for serverless Python execution, the webhook layer that turns events into work.
- Python for the execution scripts, deterministic and individually testable.
- Claude at the orchestration layer, reading directives and routing work.
Tool-by-tool reasoning is at /stack.
05 The engineering decisions that made it reliable
One database, one truth. Every script writes its results back to the same store the dashboard reads. There is no synchronization step, because there is nothing to synchronize.
Scripts over prompts for anything that must be right. Financial math, API pagination, retry logic: deterministic code. The model layer writes strategy and prose, never ledger rows.
Every run leaves a trail. Workflow executions are recorded with status and logs, so when something fails, the question is "which run, which line" rather than "what happened".
Strict isolation from client data. This system runs the agency. Client systems run on their own isolated database instances, and the two never mix. The internal dashboard can see that a client engagement is healthy; it cannot see into the client's data, by construction.
06 What running it taught us
The build paid for itself in assembly work nobody does anymore, but the durable lesson is about ownership. Because the behavior lives in plain-text directives, changing how the agency operates is an edit, not a project. New lead source? A directive and a script. New report? A surface over data that already exists.
That experience is why every client engagement includes education on the same architecture. We are not teaching clients a theory. We are teaching them the operating system we run on, with our own P&L as the proof of confidence. The pattern is productized as the operations hub.
07 Related work
The sales side of this system is the reply-to-deck pipeline, and its memory is the knowledge vault. For the client-facing version of an operations build with measured outcomes, see the PE deal desk case. To find out what an operations layer would reclaim in your business, start with the free audit.