A platform with thousands of third-party products. And a five-figure monthly agency bill.
Give InKind builds a unified platform for community support: when someone is going through a hard time, friends and family coordinate meals, errands, and gift cards through a single interface. The product depth comes from third-party integrations across gift cards, retailers, and services. Each integration needs a product listing that's accurate, on-brand, and structured for search.
The catalog grew faster than the in-house team could write for it. Lauren's team was running thousands of product listings through external agencies, plus landing pages, plus marketing copy. Monthly agency spend ran into five figures. The team got the work done, but every dollar that went out the door came back as content that had to be reviewed and tone-corrected anyway, because external agency writers couldn't fully match the Give InKind voice.
“We've saved thousands of dollars per month by not using external agencies and starting to use Writesonic. We have written it for the landing page copy. I really appreciated that.”
Laura Malcolm, Founder of Give InKindGeneric AI tools weren't equipped for the catalog.
Give InKind's product listings aren't generic. Each one describes a third-party offering (a gift card brand, a service provider, a meal kit) with the right tone for someone navigating a difficult life moment. Generic AI writing tools produced output that was either too commercial or tonally off for the support context.
The volume requirement was also real. 16,000 listings is catalog-scale. Single-prompt AI tools didn't have the architecture to keep brand voice and tonal sensitivity consistent across that volume.
What changed with a brand-trained pipeline that handles catalog scale.
Writesonic's content engine runs each piece (product listing or landing page) through a multi-stage pipeline. Research against existing source material and brand context. Outline generation tuned to the content format (listing vs landing page vs marketing copy). Brand-voice training on Give InKind's existing high-quality content to capture the support-platform tone. Multiple expert-role review passes for brand safety, factual accuracy, and tonal fit. Quality gates that catch problems before publish.
For Lauren's team, the catalog operation moved from agency-managed to in-house validated. The pipeline produced output that matched the brand, which meant the team validated rather than rewrote.
“We had an offshore team crank through 16,000 product listings. Which is why we have used 3.5 million words.”
The 3.5 million words isn't a vanity metric. It's the throughput signal that the team kept catalog updates current rather than letting third-party integrations sit with outdated descriptions.
16,000 listings. 3.5M words. Five-figure monthly spend, gone.
• Product listings: bottlenecked by agency throughput → 16,000 produced
• Total words generated: limited by agency budget → 3.5 million across formats
• Monthly agency spend: $10,000s → eliminated
• Content scope in-house: limited → product listings + landing pages
The in-housing of landing-page copy alongside product listings is the operational shift that compounded. Most teams who move to AI content move only one format and keep agencies on for the rest. Give InKind moved both onto the same pipeline.
What the pipeline does that prompt tools don't.
For a platform managing thousands of third-party product listings plus landing-page work, three pipeline behaviors carry the weight:
• Brand-voice training that holds across formats. The pipeline produces consistent voice across product listings, landing pages, and marketing copy.
• Catalog-scale quality consistency. Quality gates and revision loops catch tonal or factual issues at every individual listing rather than degrading as volume scales.
• Format-aware research grounding. Each piece is grounded in the actual third-party offering or landing-page intent rather than in a model's training data.
For Give InKind, that translates into a content operation that matches the agency's previous output at a fraction of the line item.
