Skip to content
ytpartners

ytpartners case study.

Yarden Tadmor · 2024 - 2025

Applied AI in post-sale execution systems

Campaign delivery ran on people catching things by hand, and anything missed at intake became rework three weeks later. I built the intake agent myself and wrote its instructions, then put routing, QA and exception handling around it, so a problem surfaces on arrival instead of three weeks later.

Back to client case example

Client snapshot
Client category
Creator economy marketing platform
Channels
YouTube, TikTok, Instagram, Podcasts
Model
Managed + platform
Constraint
Rework and exceptions
Lever
Routing + QA automation
Outcome
Higher throughput

Executive summary

The operational ceiling came from manual validation, coordination overhead, and exception-driven rework. The solution was not a single automation. It was a system: intake validation, tier-based routing, embedded QA, and exception detection tied to owners and escalation rules.

What I built

AI brief conversion and validation

Brand briefs arrived in whatever shape the brand sent them. I wrote an agent that converts each one into the agency's own brief format: an eight-section template covering integration requirements, talking points, on-screen disclaimers, call to action, and product features.

Two rules in it matter more than the conversion. It is forbidden to supply any fact the source brief does not contain, and anything the source is missing is marked MISSING in place and listed again at the end. A gap in a brand's brief becomes visible on day one instead of becoming a reshoot. The agent then re-reads its own output against the source, checks that nothing was invented and nothing was dropped, and prints the checklist it used.

  • Eight-section output format, consistent across every brand
  • No fact may be added that the source brief does not contain
  • Missing inputs flagged in place and collected at the end
  • Self-check against the source before output, with the checklist shown

Routing and escalation logic

  • Tier-aware routing rules that match effort to economic value
  • Automation-first for the long tail, exception queues for edge cases
  • Escalation triggers for senior review where needed

AI-assisted QA and exception detection

  • Checks embedded into execution flow to reduce rework
  • Exception detection prompts before launch-critical deadlines
  • Repeatable validation patterns for consistency

Instrumentation and reporting

  • Cycle time, exception rate, rework rate, on-time launch rate
  • Cost-to-serve signals used to refine routing rules
  • Operational feedback loop to drive adoption and iteration

What changed

  • Intake became structured and validated rather than ad hoc
  • Work routing aligned to tier value and complexity
  • QA moved earlier to prevent downstream failures
  • Exceptions became detectable signals, not surprise fire drills

Assets delivered

  • Templates, rules, and validation logic for intake
  • Tier routing rules, owners, SLAs, and escalation paths
  • QA checklist and embedded checkpoints
  • Exception taxonomy and tracking logic

Outcomes

  • Lower rework load and fewer late-stage failures
  • Higher throughput without linear headcount growth
  • Better on-time launch predictability
  • Improved renewal safety through delivery consistency

Where AI mattered most

  • Validation and completeness checks at intake
  • Tier classification and routing logic
  • QA checks tied to requirements and compliance
  • Exception pattern detection and escalation triggers

Related case studies