AI Content Engine for a Consulting Firm

The Challenge

A management consulting firm with seven consultants had deep domain expertise but almost no market visibility. Their practitioners had decades of combined experience competing against firms like Accenture and Deloitte, but potential clients searching for help were finding competitors who published consistently, not the firm with the best people.

The obvious answer was AI. The immediate pushback was that AI content is garbage. And at that point, they were right. The firm had tried generic AI tools and the output was exactly what everyone feared: bland, formulaic copy that sounded like it could have come from any company in any industry. It didn't sound like their people. It didn't reflect how they think. No one wanted their name on it.

So the firm was stuck. Seven consultants with real things to say, no time to write, and a tool that produced content nobody trusted.

Key pain points:

• The consultants strongest at client delivery had the least time to write
• Previous attempts with freelance writers couldn't capture authentic voices
• Generic AI tools produced bland, interchangeable copy that eroded trust in the whole approach
• Internal skepticism was high. Leadership and consultants alike assumed AI content meant low-quality content.

The Approach

The problem wasn't AI. It was how AI was being used. Generic tools write generic content because they have no idea who they're writing for or what's worth saying. The fix was building a system that understands each person's voice and produces content that's genuinely interesting to read, not just technically accurate.

What we built:

• Individual voice profiles for each consultant, built by analyzing their existing writing. How they structure arguments, which words they reach for, what they'd never say. Not a tone selector. A detailed model of how each person actually communicates.
• A source content library indexing every existing article, post, and presentation as style references for voice-matched generation
• AI slop detection that catches the patterns that make content feel machine-generated: filler phrases, unnatural transitions, generic conclusions, over-hedging. If a draft wasn't substantive and interesting, it didn't move forward.
• An editorial workflow with role-based permissions, version tracking, and an approval pipeline so nothing publishes without human sign-off

The system was designed to convert the skeptics. Every draft had to clear two bars: does it sound like the person whose name is on it, and would you actually want to read it? If either answer was no, it went back through the system, not to a human to rewrite from scratch.

The Results

The consultants who were most skeptical became the most active users. Once they saw drafts that actually sounded like them, the resistance disappeared. The conversation shifted from "AI can't write for me" to "can it do my presentation notes too?"

Consultants who had never published started publishing regularly. The marketing coordinator's role shifted from chasing people for content to coordinating a reliable content calendar. Within the first month, organic search traffic went up. On LinkedIn, practitioners started getting direct messages that turned into discovery calls. Prospects began referencing specific articles in initial conversations. Content was generating inbound leads for the first time.

What's Next

The client is exploring extensions including client-facing content generation, training material pipelines, and using the same system for email campaigns, webinar scripts, and podcast notes. The voice profiles built during this engagement carry forward to every new use case without starting over.