When AI Companions Reach Their Expiration Date
1. The Setup
We launched a social campaign around an ongoing sensitive topic. The stakes were high: we needed to create meaningful conversation, not just chase vanity engagement. With a content series ahead and a creator partnership to manage, I built a project companion AI to shoulder the strategic thinking and keep us anchored to the brief through all the inevitable drift that happens when teams start riffing on ideas.
2. My Approach + AI Role
I prototyped a project companion specifically designed to hold the strategy and creative anchors. Think of it as a guardrail system: as the team explored creative concepts and messaging angles, the companion could reference back to the core problem we were solving and our intentional direction. It wasn’t about generating content. It was about being the voice in the room that asks, “Does this still serve the brief?” The companion held the strategic foundation so the team could focus on creative problem-solving without losing sight of purpose.
2.5. The Unexpected Dividends
What I didn’t anticipate was how the companion would solve onboarding friction. Projects always generate ad hoc asks. You need to bring in someone for social listening, or production support, or design execution. Normally they’d have to dig through decks littered with internal notes, external feedback, and endless rounds of review comments stacked on top of each other. Instead, we used the companion to generate:
- Social listening briefs
- Production briefs
- Design briefs Each one grounded in the strategic foundation we’d already established. New teammates could get up to speed without archaeological digs through version histories. The other surprise? We started copying and pasting client questions directly into the companion to get its perspective. It gave the team confidence to respond because all that early grounding work through strategy and creative was now right there when we need it. The companion became both a briefing tool and a confidence layer.
3. What Actually Happened
The companion was essential during planning and creative development. Then we moved into production, and something shifted. The account lead and I realized the companion hadn’t been updated in over a month. Why? Because once we were working directly with the creator at higher frequency, the team naturally relied on the partnership itself for direction and those minor detailed tweaks that come with execution. The companion had done its job and then quietly became obsolete.
4. The Real Insight
Project companions have a shelf life tied to project phase. Their value peaks during strategic foundation-setting and creative development when teams need consistent anchoring against drift. Once you hit execution mode, where it’s less about “what should we do” and more about “how do we make this happen,” the dynamics change completely. Execution demands high-frequency, visual coordination (timelines, asset trackers, production schedules) and direct collaboration with partners. The strategic questions have been answered. Now teams need to see the plan come together, not debate whether the plan is right. I may have discovered that AI companions work best as phase-specific tools, not project-long partners. But that raises another question: maybe there’s a different approach to context updates that doesn’t bloat into massive overhead. Something lightweight enough to keep the companion relevant without becoming another thing to maintain.
5. Try This If…
If you’re launching complex campaigns with multiple stakeholders: Build a project companion for the planning and creative phases. Let it hold your strategic anchors and keep teams honest about purpose. But don’t force it into execution. When you shift to production, acknowledge the companion served its purpose and let visual tools (Gantt charts, asset matrices, production boards) take over. If you’re managing creator partnerships: Use the companion until the relationship with the creator becomes the primary coordination mechanism. The high-frequency exchange, idea sharing, and refinement of what you’re about to do together is where the real value lives when you’re close to the work. But keep your Spidey senses active. If something feels off-brief, bring the companion back in to ask open-ended, thoughtful, challenging questions about the current set of assumptions and the direction you’re about to head into.
6. Systems Lens
This reinforces something I’ve been circling around: AI augmentation isn’t about replacing workflows end-to-end. It’s about identifying the specific friction points where AI can provide leverage, then stepping back when direct collaboration or visual coordination become the better tool. Most people think about AI as “set it and forget it” automation. But what if the real skill is knowing when to sunset the AI because the system has evolved past needing it? The companion didn’t fail. It succeeded so well that the team internalized its function and moved on. This also raises a bigger question I need to investigate: What would a project companion look like during execution? Could there be a version that shifts from strategic anchor to real-time production support, surfacing blockers or suggesting resource reallocation based on progress? Or is execution fundamentally too visual and coordination-heavy for AI to add value beyond task management? The answer probably depends on campaign complexity. For a single-platform content series, probably not. For a multi-platform integrated campaign with different messaging needs across touchpoints, maybe. That’s worth testing next time I’m in that context.
Next Experiment: Figure out what high-complexity campaign execution actually needs (asset tracking, cross-team dependencies, messaging consistency checks) and test whether an AI companion could serve those functions better than our current visual tools, or if it would just be noise.
Wish me luck.