From Burnout to Breakthrough: 7x Your Agency with AI

Situating AI in Context: Perspective for the Overwhelmed

Before getting into the specifics of how Designbox uses AI, Thomas took time to address the anxiety he sees many agency owners and freelancers experiencing around AI adoption.

He acknowledged that the volume of new tools, updates, and approaches can feel genuinely overwhelming. Every week something new arrives. It is easy to feel perpetually behind.

His reframe: ChatGPT only launched in November 2022. That is roughly three and a half years ago. He drew a parallel to the early internet, noting that the first website went live in 1991 — he mentioned the link is still live and shared it as a curiosity — and yet nobody was building e-commerce websites in 1995. Broad commercial adoption took years. The current moment in AI, by that measure, is still very early, and being early is an advantage, not a liability.

He reinforced this by noting that agency owners and freelancers do not need to become AI developers. There are software companies already doing the hard technical work, adapting their platforms at speed, and making AI increasingly accessible. He gave Atarim as an example of a company he has observed staying ahead of the curve on AI integration. His point was clear: your job is not to build everything from scratch. Your job is to understand where AI fits into your business and to start applying it to the right problems.

His recommended starting point: identify the single biggest pain or bottleneck in your business right now and focus on that one thing first. Do not try to do everything at once.

The Foundation: Systems and Processes Before AI

Thomas used an AI-generated image of himself wearing multiple hats as a metaphor for the classic overwhelmed agency owner. He was explicit that Designbox went through exactly that phase. He was account manager, HR, director, and everything else simultaneously, which made scaling feel impossible.

The solution they identified was building robust systems and processes so that others could take over functions. About five or six years ago, while he was a mentee at Agency Mavericks (working with a coach named Johnny Flash), he and his team decided to build over one thousand SOPs in a single year.

He described this initiative with candour and humour. They created SOPs on everything, including SOPs about creating SOPs. They had an SOP on how to answer the phone. They had four separate SOPs covering how to create a Google Analytics 4 account, and he believes only two of them were correct.

The outcome: despite the volume of documentation produced, almost none of it was useful.

The reason was that the team started with tasks rather than strategy. They jumped straight to micro-level instructions without first mapping the journey those instructions were meant to serve. Thomas made an analogy: you would not expect your accountant to have an SOP for how to use Google Sheets. You would expect them to already know it. The SOPs that mattered were the ones governing how the business moves a client from first contact to a delivered outcome — the strategy and journey layer, not the basic tool-operation layer.

The lesson he drew: you cannot plug AI into chaos. If your processes are a mess, AI will produce a bigger mess faster. The foundation has to come first.

The Five-Step Process for Fixing Broken Systems

To explain how he approached cleaning up their processes and journeys, Thomas referenced Elon Musk's five-step method for fixing a broken system, which he said he started applying to Designbox's user journeys and SOPs. He noted, without elaborating extensively, that Musk runs some of the most complex organisations in the world, making him a reasonable reference point for systematic thinking about process.

The five steps he described: 1. Question every task. Look at each step in a process and ask whether it is actually needed. If no client is asking for it, if removing it does not break the outcome, it may not be necessary. Delete what does not add value.

2. Simplify. After questioning, look for ways to reduce complexity. A 24-step Loom video on how to log into a website is almost certainly overcomplicating things. Remove friction wherever possible.

3. Accelerate. Once the process is as simple as it can be, find ways to do it faster.

4. Automate. Only after the above three steps should you look at automation. If you automate a broken or unnecessarily complex process, you automate the problems inside it.

5. (Implied) Apply AI. AI sits at the end of this chain. It is the accelerant on top of a system that already works.

Thomas was direct about the common mistake: most agency owners get excited and start at step four or five. They try to automate or use AI before they have done the harder work of questioning, simplifying, and streamlining. This is where the investment fails to deliver.

If you automate a broken or unnecessarily complex process, you automate the problems inside it.

Mapping the Customer Journey First

Thomas briefly showed a screenshot from Designbox's Whimsical workspace to illustrate what a properly mapped journey looks like. He was careful to note that what he was showing was only a slice of their sales journey — specifically, the initial contact and inquiry handling stage.

Within that stage, they have mapped:

• How to handle inquiries depending on their source
• How to qualify an incoming lead
• How to handle a sales call
• What to do after a sales call He emphasised that this is how SOPs should be built: derive them from the journey, not the other way around. The journey tells you what SOPs you need. If you start with SOPs before you understand the journey, you get the kind of chaotic documentation exercise Designbox experienced in their early years.

The reason they started with sales specifically was pragmatic: Thomas could not do all the selling himself, and he needed a predictable outcome from the sales process. Mapping it first was what made it possible to hand it to others effectively.

If you start with SOPs before you understand the journey, you get the kind of chaotic documentation exercise Designbox experienced in their early years.

Where AI Fits in the Agency Workflow

Thomas walked through the range of tasks that eat up time inside agencies of every size — whether a solo freelancer or a 50-plus-person firm. He listed examples including: replying to inquiries, writing proposals, onboarding new clients, following up, and updating strategies.

These tasks, he said, follow repeatable patterns, and patterns are exactly where AI can assist.

He was careful to frame AI as an assistant alongside the team, not a replacement for it. His framing: imagine giving every team member multiple AI assistants. How much more productive would they be? That is where AI shines — not in replacing competent people, but in amplifying what they can do.

He mentioned that he shared a list of specific examples of how Designbox uses AI agents across different business functions, with the intent to share that resource with attendees after the session.

He also pointed out that some attendees may already be using AI tools without fully recognising them as such. Using Fathom or Fireflies to take meeting notes, for example, is already AI integration. The question is whether that output is being connected meaningfully into a broader process.

The Core Problem Designbox Set Out to Solve: Sales Cycle Time

Thomas posed a direct question to the audience: how long does it take your agency to move from an incoming inquiry to a delivered proposal or strategy?

He then walked through what the Designbox process looked like before their AI system: Inquiry comes in Team qualifies the lead Book a sales call One-hour phone call with the prospect, plus approximately 15 to 20 minutes of preparation before the call Create a strategy or proposal after the call Hold a follow-up meeting to walk through the strategy or proposal Send it to the client Follow up Send the invoice Begin onboarding Their calculation: end-to-end, when done properly and with appropriate customisation, this process took approximately seven hours per prospect. For an agency handling multiple 1.

inquiries, this is a significant time and resource drain — especially if a proportion of those prospects do not convert.

This became the target for AI intervention.

The Designbox AI System: Live Demonstration

Thomas demonstrated a custom internal platform — referred to during the session as a portal or agency portal — that Designbox built to handle the complete sales process from inquiry through to signed proposal and automated onboarding. The platform covers proposals, the sales pipeline, client onboarding, audits for lead generation, and more.

He noted the platform was built iteratively: tested, broken, repaired, and updated repeatedly over time before reaching a state that produced consistently good outputs.

The triage assistant: when an inquiry arrives, the team uses a triage assistant built into the platform. It asks the team member a series of approximately ten questions based on the inquiry context. The output is a complete meeting agenda for the team member handling the call, generated ahead of the meeting. This includes:

• A structured list of questions to ask the prospect, covering the key information needed to build a strategy.
• Guidance on being flexible, since prospects often answer questions with questions and the conversation should remain natural.
• A brief on red flags to watch for and how to adapt.
• A closing script, including guidance on handling objections and hesitations.

Thomas emphasised that this preparation does not mean the team member has to ask every question on the list rigidly. The goal is to ensure that at least one or two questions from each section are covered, so the information needed for the strategy is gathered.

The output of each call can then be added to the CRM/pipeline.

The Digital Roadmap Builder After the meeting, the team builds what Designbox calls a digital roadmap — their term for a digital marketing strategy. Thomas demonstrated this component live.

The key innovation is that the system has been trained on everything Designbox does: their specific service plans, how they price them, what they deliver at each tier, and how they operate. This is the crucial differentiator from using a general AI tool like ChatGPT or Claude directly. When you prompt a general AI tool to build a strategy, it does not know what your agency delivers. It may produce a coherent-sounding plan that has nothing to do with

your actual service offering. Designbox's platform only works within the services and plans they have defined, so the output is always grounded in what they can actually deliver.

The process within the roadmap builder: The team member selects the appropriate service plans for the prospect (for example, the basic SEO plan and the boost email marketing plan).

They paste in the meeting transcript. Thomas mentioned they are currently integrating an API with Fathom AI so this step becomes automatic rather than requiring copy-paste.

The system uses the inquiry data from the triage assistant plus the transcript to generate a personalised introduction letter to the prospect.

It then identifies growth opportunities specific to that prospect, grounded in the services being proposed.

It generates relevant KPIs, with the caveat that this works best when the team has real benchmark data to draw on.

It builds out a month-by-month plan for the services selected, drawing from predefined service descriptions so the content is consistent and accurate.

It suggests campaign activities month-by-month, and the team can refresh suggestions or add custom ones if the initial output is not right.

The full growth roadmap is generated and can be shared with the client.

Thomas showed how the final document presents growth opportunities, expected outcomes month-by-month, and a clear breakdown of what is included in each service plan. When the prospect is ready to move forward, they click a "ready to go" button that triggers the next stage: proposal creation, digital signing, and automated onboarding.

Time saved: Thomas estimated that while a human team member might still want to spend around an hour reviewing and refining the output to ensure quality, this is dramatically faster than the previous seven-hour average. He was straightforward that spending an hour on review is still appropriate — but one hour versus seven is a meaningful difference.

Expansion: Building Beyond the Initial Problem

Once Designbox had built the sales and strategy system, Thomas said he "got the bug."

Solving one problem revealed others, and the team continued building. Additional features and systems they have built or are building include: Automated client reports Automated onboarding sequences to maintain client engagement immediately after signing (avoiding the cold drop-off that can happen post-sale) A system to collect content from clients so that projects keep moving without delays 1.

Personalised, well-designed proposals (replacing either generic templates or fully custom builds done from scratch each time) He noted that they never set out to build software. The platform emerged from the specific problems they were trying to solve for their own agency. It was problem-first, solution-second — not "let's build a SaaS product" as a starting point.

Agencio: Commercialising What They Built

Thomas announced that Designbox is in the process of launching Agencio, a SaaS product built from the systems and tools developed for their own agency. He said there has been demand from people he has coached through Agency Mavericks and others in his network who want access to the platform rather than having to build everything themselves.

He described a QR code displayed during the session that leads to a waiting list. The platform was described as being "a few months" from launch at the time of the session. The stated ambition is to cover all the pain points within a typical agency business, particularly around sales, in a single integrated platform.

He acknowledged that there are other tools available that address individual parts of the business, but framed Agencio as an attempt to cover the full spectrum in one place.

Outlook: Where Agencies Are Headed in Two Years

The host, Andrew Palmer, closed the Q&A by asking Thomas where he thinks website agencies and creative agencies will be in two years.

Thomas's answer was optimistic. He believes the current pain points — the ones that feel significant and frustrating now — will be largely forgotten, in the same way that the disappearance of Dreamweaver and Flash feels like ancient history. He specifically called out content collection from clients as a current source of friction that he expects AI to largely resolve, making projects faster and smoother to run.

His broader view: agencies will be able to deliver significantly more value for clients in less time. Things that previously took months will take weeks or days. He expressed genuine excitement about where the industry is going rather than anxiety about what AI might displace.

About the speaker Thomas Amos Founder of Designbox and Agencio, Coach at Agency Mavericks

Thomas Amos is founder of Designbox and Agencio, and a coach at Agency Mavericks. He works with agency owners on process, automation, and using AI to remove bottlenecks without losing the human judgment that clients pay for.