🎮 The Next Input — Issue #059

In partnership with

⚡ The Briefing — 60 sec

• OpenAI taps Apple supplier to make its AI device. I see what you’re doing, Jony.

• Silicon Valley bets big on environments to train AI agents. Another bet. Is this even news at this point?

• The missing piece of the $2T AI market is human psychology. Couldn’t agree more.

🛠️ The Playbook — AI Device Companion Dashboard

Mission Build a central “mission control” dashboard for AI devices (smartphones, wearables, desktops) that logs usage, tracks costs, and offers explainability for every interaction.
Difficulty Advanced | Build time 2–4 hours (pilot)
ROI Saves ops teams ≈ 12–15 h/month and prevents runaway API bills or opaque usage patterns.

0) Why This Matters

As OpenAI and Apple suppliers push hardware into the mix, companies will soon deploy fleets of AI-enabled devices. Without observability, you’re blind to:

• Costs per employee/device.

• Types of tasks automated.

• Compliance risk from unreviewed outputs.

1) Architecture

2) Workflow

1. Capture

   • Each device agent records {user_id, timestamp, task_text, tokens_used, output_length}.

2. Stream & Store

   • Logs flow into PubSub → Supabase/BigQuery.

3. Categorise

   • LLM tags: “creative”, “research”, “ops”, “customer-facing”, “sensitive”.

4. Score Risk

   • Rulebook: if text contains PII/client terms → flag high-risk.

5. Dashboard Update

   • Daily refresh: cost/user, top task categories, flagged risks.

6. Alert

   • If spend >$X/day or flagged tasks >Y% → send Slack ping.

3) Prompts

Categoriser Prompt

SYSTEM: You are an AI auditor.
TASK: Categorise this device interaction into one of:
["Creative","Research","Ops","Customer-Facing","Sensitive"].
Also output {risk_score 0-1, reason, estimated_cost}.
Return JSON only.

4) Guardrails

• PII Masking – Strip account numbers, client names before LLM processing.

• Token Budget – Enforce per-user monthly caps.

• Role Segregation – Sales tasks vs. Engineering tasks tracked separately.

• Explainability – Each flagged task links back to source log + rationale.

5) Pilot Rollout — 2 Hours

1. Deploy lightweight Chrome extension to 5 pilot users.

2. Capture queries + outputs into Google Sheets via Zapier.

3. Run Claude classification → append tags + costs.

4. Build Looker dashboard with top-10 tasks + spend.

5. Iterate rules before scaling org-wide.

6) Metrics That Matter

• Cost per user vs. baseline.

• % tasks flagged “sensitive.”

• Hours saved → estimate by task category (creative vs ops).

• Anomaly frequency (sudden spikes in cost/usage).

🎯 The Arsenal — Tools & Prompts

Summarise logs into: 
- Total spend 
- Top 5 task categories 
- % sensitive tasks 
- Notable anomalies 
Output in bullet digest for Slack.

💡 Free Office Hours

Want a device companion dashboard before your AI hardware rollout?
Book a free 15-minute Office Hours slot—no sales pitch, just workflows solved.

→ Grab a slot: https://calendly.com/aaron-cylentis/the-next-input-office-hours

A free newsletter with the marketing ideas you need

The best marketing ideas come from marketers who live it.

That’s what this newsletter delivers.

The Marketing Millennials is a look inside what’s working right now for other marketers. No theory. No fluff. Just real insights and ideas you can actually use—from marketers who’ve been there, done that, and are sharing the playbook.

Every newsletter is written by Daniel Murray, a marketer obsessed with what goes into great marketing. Expect fresh takes, hot topics, and the kind of stuff you’ll want to steal for your next campaign.

Because marketing shouldn’t feel like guesswork. And you shouldn’t have to dig for the good stuff.

Subscribe Free

🕹️ Game Over

Ship one device log today—tomorrow your CFO and CISO will thank you.
Share your win; you could headline Issue #060.

— Aaron
Automating the boring. Amplifying the brilliant.

Forwarded this? Subscribe here