The Agency Hex: Team Workflows Under the Constraint
The solo hex is hard in one direction — you have to be honest with yourself about which tools you actually use. The agency hex is hard in five directions at once, because every person on the team has a different tool they consider non-negotiable, and nobody wants to be the one whose favorite gets cut. A five-person content agency ran the hex as a team exercise and discovered something that made the conversation uncomfortable: the tool preferences they were defending had less to do with productivity and more to do with identity.
This is the case study of what happens when the hex moves from personal discipline to organizational decision.
The Problem
Five people. Fifteen-plus active subscriptions. Zero workflow compatibility.
The agency produces blog posts, social media content, and email campaigns for twelve clients. The team includes two writers, a designer, a strategist, and an account manager. Each person had adopted their own AI tools over the previous year, independently, without any coordination. The result was a tool environment that looked less like a stack and more like a drawer full of chargers — each one worked fine for the device it came with, and none of them fit anything else.
Writer A used Claude for drafts and Grammarly for editing. Writer B used ChatGPT for drafts and Hemingway Editor for editing. The designer used Midjourney for concepts and Adobe Firefly for client-ready assets. The strategist used ChatGPT for research, a separate competitive analysis tool, and Notion AI for organizing insights. The account manager used Claude for client communications and a third-party meeting summary tool for call notes. Nobody's workflow touched anybody else's. When Writer A went on vacation, Writer B couldn't pick up their client's work without spending an hour figuring out how the Claude project contexts were set up — assuming they had Claude access at all, which they didn't, because each person was paying for their own subscriptions on their own credit cards.
The subscription cost was distributed across personal expenses and the agency's overhead in a way that nobody had fully tallied. When the agency owner finally added it up — combining the agency-paid tools with the personal subscriptions the team was expensing — the number was $890/month [VERIFY]. For a five-person agency billing $40K-60K/month in client work, the tool spend wasn't going to bankrupt anyone. But the workflow incompatibility was costing something harder to measure: the inability to share work, cover for each other, or build institutional knowledge that survived any single person leaving.
Applying the Hex to a Team
The hex constraint for an individual is "six tools, each earning its slot." The hex constraint for a team adds a second dimension: the tools have to be shared. Not "everyone has access to the same tools" — that's just a licensing decision. Shared means the tools are configured the same way, the workflows are documented, and any team member can sit down at any client's work and start producing within fifteen minutes.
The agency held what they later called "the tool argument" — a three-hour meeting where each person presented their tools and made the case for keeping them. The meeting was supposed to take one hour. It didn't, because tools are personal in a way that office furniture and email clients are not. When Writer A said "Claude is better for long-form," Writer B heard "your tool is worse." When the designer said "Midjourney produces higher quality than anything else," the strategist heard "my visual suggestions aren't good enough." The tool argument was really an identity argument. Every person had built competence in their specific tools, and competence feels like an asset — because it is one, right up until the organization needs interoperability more than it needs individual optimization.
The resolution came from the constraint itself. The hex doesn't ask "which tool is best." It asks "which tools produce client deliverables." That question has an answer that doesn't require anyone to admit their favorite tool is inferior. It just requires everyone to agree on which tools the agency — not any individual — will use.
The Six That Made It
One LLM: Claude, on a team plan. This was the most contested decision. Writer B had years of ChatGPT muscle memory and a library of custom instructions that represented genuine accumulated skill. The decision ultimately came down to one practical factor — Claude's Projects feature allowed the agency to create persistent client contexts that any team member could access. ChatGPT's custom GPTs offered something similar, but the agency tested both and found that Claude's project structure mapped more naturally to "one project per client" in a way the whole team could navigate. Writer B spent a week migrating and reported — honestly — that the adjustment was frustrating but survivable.
One image tool: Midjourney, on a shared account. The designer kept their tool. Adobe Firefly got cut — not because it was worse, but because maintaining two image generation pipelines for different "quality tiers" meant the designer was the only person who knew which tool to use for which client. One tool meant anyone could generate an image if the designer was unavailable.
One project management tool: Notion, stripped down. The agency had been running Notion alongside Asana alongside a shared Google Sheet that the account manager maintained as the "real" source of truth. The hex killed Asana and the Google Sheet. Notion became the single source. Every client got a database entry. Every deliverable got a row. The Notion AI features — which the strategist had been using independently — became available to the whole team, though in practice the team used them sparingly.
One design tool: Canva Pro, on a team plan. This wasn't contested. Every team member occasionally needed to resize an image or adjust a social graphic, and Canva's interface required zero design training.
One communication tool: Slack, with all AI integrations disabled. The agency had been paying for Slack AI, a meeting summary bot, and a third-party Slack plugin that used AI to prioritize messages. All three were removed. The team agreed — reluctantly at first, then with visible relief — that Slack's job was to carry messages, not to summarize, prioritize, or interpret them. The meeting summaries were replaced by a simple template that the account manager filled out manually in two minutes per meeting. The AI message prioritization was replaced by a channel structure that put client-urgent items in dedicated channels.
One analytics tool: Google Search Console plus each client's native analytics. The two third-party analytics platforms the strategist had been using were consolidated into direct access to client dashboards. Less polished reports. Same data. No monthly fee.
Six tools. Agency-wide cost: $340/month [VERIFY], down from $890. Every tool shared. Every workflow documented. Any team member can pick up any client.
What the Team Resisted
The productivity dip was real and lasted about three weeks. Writer B's output dropped by roughly 20% during the first week of the Claude migration — not because Claude was slower, but because the muscle memory hadn't transferred and every task took slightly longer than it used to. The designer complained that Midjourney alone couldn't handle every client's visual needs and lobbied for a second image tool for two weeks before acknowledging that the issue was more about aesthetic preference than capability.
The strategist lost the competitive analysis tool and genuinely felt the gap. The tool had automated competitor content monitoring in a way that Claude couldn't replicate directly. The workaround — a Claude prompt that analyzed competitor URLs fed to it manually — produced 80% of the same insights in 120% of the time. The strategist called this "a downgrade" and was technically correct. The agency accepted the tradeoff because the 20% of insights being lost weren't showing up in client deliverables. They were showing up in strategy decks that impressed clients without changing the actual content plan.
The account manager had the easiest transition. Their tools were mostly communication-adjacent, and the move from AI-enhanced communication to plain communication required nothing more than typing slightly more words per message.
The Workflow Compatibility Win
Six months after the hex, the agency experienced the event that validated the entire exercise: Writer A left for a full-time position.
In the pre-hex environment, this would have been a crisis. Writer A's Claude projects, Grammarly settings, client voice notes, and personal workflow documentation would have left with them — or more accurately, would have never existed in a transferable form. The agency would have spent two to four weeks reconstructing client contexts for whoever replaced Writer A.
In the post-hex environment, the transition took two days. Writer A's clients lived in shared Claude Projects that the whole team could access. The Notion database had every client's deliverable history, style notes, and feedback patterns. The new writer — Writer C — opened the project, read the context, produced a first draft on day two, and received client approval with minor edits. The client didn't know the writer had changed. The workflow was the same. The tools were the same. The institutional knowledge lived in the tools, not in the person.
That's the argument for the team hex that no individual productivity metric can make. Individual tool optimization produces faster individuals. Shared tool constraint produces a resilient organization. The agency doesn't need each person to be their fastest. It needs each person to be replaceable — not in the dehumanizing sense, but in the operational sense that the work survives any single departure.
What They Would Do Differently
The agency owner identified two things they'd change about the hex process.
First: run the tool argument before announcing the constraint. The team felt ambushed. The owner introduced the hex concept and the tool consolidation in the same meeting, which meant the team was processing a new framework and defending their tools simultaneously. A better approach — which the owner now recommends to other agencies — is to introduce the hex as a thought exercise first. "If we could only use six tools, which six?" Let the team sit with that question for a week before making it real. The emotional processing happens before the decision, not during it.
Second: give each team member one "personal tool" slot outside the shared six. The agency eventually did this at the three-month mark, after the strategist's frustration with losing the competitive analysis tool threatened to become a retention issue. Each person gets one additional tool that they pay for personally and use for their own workflow. It doesn't need to be shared, it doesn't need to be documented, and it doesn't count against the agency's six. This release valve preserved individual autonomy while maintaining organizational standardization. In practice, two of the five team members use their personal slot. The other three discovered they didn't actually want an extra tool once the pressure to defend one was removed.
The agency hex isn't about making a team use fewer tools. It's about making a team use the same tools — and discovering, in the process, that the diversity of tools was solving a problem that didn't need solving while creating a problem that nobody had named.
This is part of CustomClanker's Hex in the Wild series — real setups from real people. Start with The Hex Explained if you haven't downloaded the constraint PDF yet.