Pod mapping is a task classification framework that tags every single task in your business as Automate, Assist, or Keep — giving you a clear, honest picture of exactly where AI fits in your operations and where it doesn’t. It’s the first thing that happens in every Zero Hire Method sprint because you can’t automate intelligently without knowing what you’re working with.
Why do you need a framework for deciding what to automate?
Because most people get it wrong without one, and getting it wrong is expensive. You either automate too aggressively — trying to force AI into tasks that genuinely need a human, which creates a mess — or too conservatively, spending three months on one automation when you could have knocked out ten quick wins in the same time.
McKinsey’s 2025 automation research found that only 5% of occupations can be fully automated, but 60% of occupations have at least 30% of activities that could be automated with existing technology (McKinsey Global Institute, “A New Future of Work,” 2025). The challenge isn’t whether AI can help your business — it obviously can. The challenge is figuring out which specific tasks to hand over, which ones need AI plus a human, and which ones you leave well alone.
That’s what pod mapping solves. It gives you a systematic way to look at every task across your entire business and make that call properly, instead of guessing or just automating whatever seems easiest.
What are the three pod mapping categories?
Every task gets exactly one tag. No “maybes,” no “we’ll figure it out later.” Updated April 2026, here’s how each category works:
Automate — AI handles it completely
These are tasks where a human adds zero value. The task is repetitive, follows clear patterns, and the output is predictable enough that you’d trust a well-built system to handle it without checking.
Real examples from service businesses: - Email sorting and triage — AI reads incoming mail, categorises by urgency and type, and routes to the right place - CRM updates after calls — AI listens to call notes and updates deal stages, contact info, and next actions automatically - Invoice data entry — AI reads invoices, pulls out the numbers, and logs them in your accounting software - Appointment reminders — AI sends personalised reminders based on your calendar, handles rescheduling - Expense categorisation — AI reads receipts, categorises spend, and matches to projects
These tasks have something in common: they’re pure pattern recognition. There’s no ambiguity, no judgment call, no relationship at stake. A human doing this work is essentially acting as a very expensive processor, and according to the Office for National Statistics, UK workers spend an average of 2.1 hours per day on tasks they describe as “routine and repetitive” (ONS, Time Use Survey, 2024). That’s over 10 hours a week of human time on work that follows rules a machine can learn.
Assist — AI does 80%, a human reviews
This is where it gets interesting because this category is where most of the value actually sits. Assist tasks are ones where AI can do the heavy lifting — the research, the drafting, the data gathering, the initial analysis — but a human needs to review the output before it goes anywhere.
Real examples: - Client proposals — AI drafts the proposal based on your templates, discovery notes, and pricing, but you review before sending - Email replies to important clients — AI drafts the response with the right tone and context, you tweak and send - Weekly reports — AI pulls the data, generates the charts, writes the summary, you check it makes sense - Content creation — AI writes the first draft of your LinkedIn post or newsletter, you edit for voice and accuracy - Client onboarding — AI prepares all the documents and communications, you handle the welcome call
“The Assist category is where I see the biggest ‘aha moment’ with business owners. They assumed these tasks were Keep — fully human — because they require judgment. But when you break them down, the judgment part is actually only 20% of the work. The other 80% is just gathering, formatting, and assembling. AI does that part beautifully.”
— Matthew Lowe, Founder, Zero Hire Method
The Assist category is also what separates good AI implementation from bad. Bad implementation tries to fully automate things that need human oversight, and the result is embarrassing emails, wrong numbers, and clients who lose trust. Good implementation uses AI to do the grunt work and puts a human at the decision point. Deloitte’s 2025 enterprise AI report found that organisations using human-in-the-loop AI systems reported 38% higher satisfaction rates than those using fully autonomous AI (Deloitte, “State of AI in the Enterprise,” 2025).
Keep — this stays human
Keep tasks are the ones where a human genuinely adds irreplaceable value. This isn’t about sentiment or resistance to change — it’s about honest assessment. Some tasks require emotional intelligence, deep relationships, creative judgment, or the kind of nuanced understanding that AI simply doesn’t have.
Real examples: - Discovery calls with prospects — reading body language, building rapport, sensing what’s unsaid - Difficult client conversations — delivering bad news, handling complaints, negotiating scope changes - Strategic planning — deciding where the business goes next, which markets to enter, which services to add - Team management — motivation, coaching, performance conversations, culture-building - Creative direction — the final call on brand voice, visual identity, positioning
Notice something about this list: these are the tasks that actually grow the business. They’re the high-value, high-impact work that you got into business to do. The problem in most service businesses is that these tasks get squeezed into the gaps between admin because there’s never enough time.
Pod mapping doesn’t just tell you what to automate — it tells you what’s worth protecting. The Keep list is your “this is where I add value” list, and everything else is up for grabs.
How do the four business engines work in pod mapping?
Every business runs on four engines, whether you’ve named them or not. Pod mapping classifies tasks within each engine separately because the automation potential varies massively across them.
Acquisition — how you get clients. Lead generation, marketing, sales outreach, proposals, follow-ups. In most service businesses, Acquisition is 40-50% automatable because so much of it is repetitive outreach and data gathering. The Chartered Institute of Marketing reports that SMEs spend an average of 20 hours per week on marketing and sales activities, much of which follows predictable patterns (CIM, “State of SME Marketing,” 2024).
Delivery — how you serve clients. The actual work you do. This varies hugely by industry, but the admin around delivery — project tracking, status updates, resource allocation, scheduling — is often more automatable than the delivery itself. A recruitment agency’s actual sourcing and interviewing stays human, but the job posting, CV parsing, and interview scheduling? Automate.
Support — how you keep clients. Check-ins, account management, issue resolution, renewals. Support is typically 30-40% automatable with AI handling routine inquiries, scheduling check-ins, and monitoring for early warning signs of churn. But the relationship work — understanding what a client actually needs, sensing when something’s off — that’s Keep.
Operations — how you run the business. Finance, HR, compliance, internal admin. This is where the highest automation potential usually lives. According to Sage, operations tasks in SMEs are 50-60% automatable with current technology because most of them are pure process: payroll processing, expense tracking, compliance filing, internal reporting (Sage, “Future of Small Business Operations,” 2025).
When Matthew maps a service business, the pod map covers all four engines in a single session — usually 2-4 hours. Every workflow, every task, every hand-off between people or systems. You walk away with a complete picture of your business with clear tags on everything.
What does a pod map actually look like?
Here’s a simplified example from a real recruitment agency (details changed):
| Engine | Task | Tag | Why |
|---|---|---|---|
| Acquisition | LinkedIn outreach to candidates | Automate | Pattern-based, high volume |
| Acquisition | Discovery call with client | Keep | Relationship, judgment |
| Acquisition | Writing job specs | Assist | AI drafts, recruiter refines |
| Delivery | CV screening | Assist | AI shortlists, recruiter reviews |
| Delivery | Candidate interviews | Keep | Human interaction essential |
| Support | Client check-in scheduling | Automate | Pure calendar logic |
| Support | Handling client complaints | Keep | Emotional intelligence needed |
| Operations | Timesheet processing | Automate | Data entry, pattern-based |
| Operations | Financial reporting | Assist | AI generates, owner reviews |
The full pod map for this business had over 80 tasks across the four engines. The split came out at 35% Automate, 28% Assist, and 37% Keep — which is pretty typical for a service business. That means 63% of their total task load could be partially or fully handled by AI.
Can you do pod mapping yourself?
You can start, and honestly you should — even a rough self-assessment is better than nothing. Grab a spreadsheet, list every task you or your team does in a typical week, and tag each one. You’ll spot the obvious wins immediately.
But there’s a reason the Zero Hire Method includes pod mapping as a coached exercise rather than a DIY template. When you’re inside your own business, you’ve got blind spots. Tasks you’ve been doing for years feel essential because you’ve always done them, not because they actually need you. Matthew has mapped dozens of businesses and brings pattern recognition that comes from seeing the same waste repeated across industries — the same admin bottlenecks in recruitment that show up in accounting that show up in hospitality.
The Harvard Business Review found that business owners overestimate the complexity of their own processes by an average of 40% compared to external assessments (HBR, “The Automation Paradox in Small Business,” 2024). You think your email workflow is nuanced and context-dependent. An experienced mapper looks at it and sees three rules and a template.
What happens after you’ve got your pod map?
The pod map creates a natural build order:
Wave 1 — Automate tasks first. These are your quick wins. They require no human oversight, so once they’re built, they’re completely off your plate. Most businesses get their Wave 1 automations running within the first two weeks of a Zero Hire Method sprint.
Wave 2 — Assist tasks next. These take longer because you’re designing the human-AI handoff points — where does AI stop and where does the human step in? Getting this right means the human review step takes minutes, not hours.
Keep tasks — left alone, but now they’ve got breathing room. Once Automate and Assist tasks are handled, the Keep tasks get the time and attention they deserve. You’re not rushing through a discovery call because you’ve got 30 emails waiting. You’re not half-listening to a client because you’re thinking about the invoices you need to process.
That’s the whole point, really. Pod mapping isn’t about removing humans from your business — it’s about making sure humans are doing human work, not acting as expensive processors for tasks that follow rules a machine can learn in an afternoon.
