01 · Case study
A digital marketing agency replaces its SaaS stack with its own systems.
A 10-12 person political advertising and ad tech agency had already installed an AI agent platform on its own. Eight weeks of architecture consulting later, the AI bill was capped, a $500/month CRM was gone, and a content and growth system was running the agency’s social presence without anyone writing a post.
Last reviewed:
Industry
Political advertising, programmatic TV / CTV / OTT
Company
~10-12 people, ~25 clients
Engagement
AI systems consulting, 8 weeks
The situation
The agency had already installed an AI agent platform and started building automations on its own, but the basics weren’t working: broken chat integrations, a news briefing that only ran once, agents that claimed to be working and produced nothing. AI usage was scattered across four different providers with no visibility into where the money was going, until a $20-25 day and 30 million tokens burned in a single session made the problem impossible to ignore.
The founder wanted to build his own internal tools rather than keep paying for a stack of SaaS tools, but needed someone who could fix the architecture, not just the symptoms.
What we did
Over eight weeks of weekly working sessions, we acted as the technical architect behind the build:
- Fixed the broken integrations that were blocking any real automation from running
- Designed a CTO-style agent architecture with task tracking that ended the “babysitting problem,” where agents reported progress on work they hadn’t actually done
- Rebuilt an overloaded, unreliable growth agent into four focused agents: one to find target accounts, one to follow them, one to engage, one to report on results
- Migrated the entire agent stack off uncapped, multi-provider API billing onto a flat-rate plan, once a scaling token bill forced the issue
- Diagnosed a silent double-posting bug that was republishing already-live content without anyone noticing
- Taught agent memory isolation to stop two internal agents from bleeding into each other’s workspace
- Designed the four-stage lead-generation architecture — signal detection, decision-maker identification, email discovery, email verification — the agency used to build its outbound sales pipeline after the engagement wrapped
The results
- AI spend capped. From $500 to $700 per month spread across four providers, down to a flat $200/month.
- A fully autonomous content pipeline. Six agents research, synthesize, write, edit, and publish to social media multiple times a day with no human writing involved.
- An autonomous growth engine. Finds and qualifies up to 500 target accounts, then follows, likes, and replies on a daily schedule with no manual outreach.
- A CRM subscription eliminated. Replaced with a custom-built CRM — deal pipeline, accounts, AI assistant — built in about two hours, cutting a $500/month line item for a 5-6 person sales team.
- A scheduling tool eliminated. Custom-built replacement, saving roughly $900/year.
- A spreadsheet tool on its way out. The ~$25/seat/month tool is being fully replaced by an internal database layer.
Add up the software line items and the AI-spend cut alone, and the agency is already saving well into five figures a year. Count the growth and content work the system now runs on its own — the kind of work that would otherwise sit on a dedicated hire — and the founder puts the combined savings at over $40,000 a year in software and staffing costs.
What this would have cost with a dev team
A comparable build — a multi-agent content and growth automation system, a browser-automation engagement engine, and a custom CRM — would typically run $40,000 to $90,000 with an outside development team over 2 to 4 months, at standard market rates for a senior contractor building reliable browser automation and a from-scratch CRM.
What it actually cost: an 8-week consulting engagement, plus AI tooling that dropped to a capped $200/month. The tools themselves were built by the founder.
I’m gonna get rid of the tools. Why use them? It just doesn’t make any more sense.
What’s next
The agency is building out the lead-generation pipeline designed during the engagement, targeting companies with in-house media-buying teams, and has floated commercializing its content system as a white-labeled $100-200/month service for other agencies.
Ready to replace your own SaaS stack?
Book a strategy callWe'll compare notes and see whether an assessment or a build makes sense for you.