Dominic Banguis, MSc - Marketing & AI Growth Systems

Home
Dominic Banguis, MSc - Marketing & AI Growth Systems

I’m Dominic Banguis, I’m an Entrepreneur, SaaS Startups & E-commerce Growth Marketer and Startup Marketing Mentor. I am also the Co-Creator of the Product / Market Fit Canvas. I am the Founder of GrowthBoxx, a fully-remote growth marketing agency that helps early stage startups with a complete outsourced marketing team to get their company running at a fraction of the cost. Over the last 15 years, I've worked with global startup founders, entrepreneurs, online startups, marketing agencies and e-commerce businesses to stand out and make a difference in their respective industries and achieve their growth goals and accelerate their companies to the next level. I specialise in Full Stack Digital Marketing such as Marketing Strategy, SEO, PPC, Social Media, Content Marketing, Email Marketing, Public Relations, Business Development, Marketing Automation, Copywriting, Analytics and Growth Hacking (using the AARRR Method). Besides being a Marketer, I am also a Freelance Web Developer with proficient skills in HTML, CSS, SQL, Php. I have also developed and managed websites using platforms such as Wordpress, Volusion, Shopify, WooCommerce.

22/07/2026

Claude can now watch your screen to learn a task, and repeat it.

Anthropic just launched "Record a Skill" inside Claude Cowork. Instead of writing long prompts or building workflows manually, you record your screen while doing a task once, narrate what you're doing, and Claude turns it into a reusable skill it can run again on its own.

Think invoice organization, file renaming, weekly reports, tracker updates. Show it once. It remembers.

This is a meaningful shift from prompt engineering to demonstration-based automation. For anyone building AI-driven systems, the barrier to teaching an AI assistant just got a lot lower.

Available now on Pro, Max, and Team plans.

17/07/2026

This week Anthropic introduced Claude for Teachers, giving verified K-12 educators in the US free access to premium Claude capabilities.

The offering includes a library of teaching skills built with learning science in mind, plus a direct connection to evidence-based curricula mapped to academic standards in all 50 states.

The rationale is simple and well-documented. Practices like differentiation, mastery-based learning, and small group instruction reliably improve outcomes, but teachers are often short on time and resources to implement them, and planning often spills into evenings.

Claude for Teachers is positioned to close that gap between what research recommends and what a teacher's actual week allows.

A few details worth noting from a systems perspective:

- Access includes both Claude Cowork and Claude Code, not just a chat interface

- It connects with existing edtech tools including ASSISTments, Brisk Teaching, Canva Education, Diffit, MagicSchool, and TeachFX

- Anthropic worked with the American Federation of - Teachers on a Gold Standard for safety and privacy in K-12 environments

- Training is off for verified teacher accounts, addressing a real trust gap

What stands out to me is the framing. This is not "AI that replaces teachers." It is AI infrastructure built around the actual job: lesson planning, differentiation, grading admin, and standards alignment. That is the same principle I apply with clients: AI systems work when they are built around a real workflow, not bolted on as a novelty.

Worth watching as OpenAI, Microsoft, and Google run the same playbook in education. The company that wins here will be the one that earns trust with the people actually doing the work.

12/07/2026

Claude just shipped a wave of updates. Here is what matters if you are running AI into your marketing and growth systems.

Claude Sonnet 5 is now the default model for Free and Pro users. It performs close to the flagship Opus 4.8 on many tasks and is priced at introductory API rates through August 31, after which standard pricing kicks in.

Cowork is moving to the cloud. Sessions and files now carry across devices, and tasks can keep running even when your device is offline. For anyone managing multiple client engagements at once, this closes a real gap.

The Microsoft 365 connector now does more than search. Claude can draft and send email, manage calendar events, and create or update OneDrive and SharePoint files, once an admin enables the permissions.

A new monthly recap feature (Settings > Reflect) shows where your Claude time actually goes: topics, peak hours, most active days. Paired with new quiet hours and break reminder settings, it is a small but useful nudge toward more intentional AI use.

Anthropic also published research on what it calls "J-Space," an internal workspace where Claude appears to hold and manipulate concepts before they show up in a response. Early days, but a meaningful data point in the interpretability conversation.

My takeaway for founders and fractional operators: the model upgrades get the headlines, but the workflow features (cross-device Cowork, connector write access, usage insights) are what actually change how AI fits into a busy operating cadence. Worth testing before the introductory pricing window closes.

What update are you most curious to put to work?

04/07/2026

Anthropic just dropped a major update that shifts the AI cost-to-performance equation.

For everyone balancing the need for top-tier AI intelligence with real-world budget constraints, Anthropic has officially released a new version of their Sonnet model.

The big takeaway?

It delivers performance that pushes incredibly close to their flagship Opus model, but at a fraction of the cost.

In the current landscape, the choice usually comes down to:

Maximum Intelligence: High cost, slower speeds, best reasoning.

Efficiency: Lower cost, fast speeds, lower complexity.

This update effectively blurs those lines. Businesses can now deploy highly sophisticated reasoning, coding, and analysis at scale without the premium price tag of a flagship model. It’s a massive win for developers and enterprises looking to optimize their AI spend without sacrificing quality.

If you've been holding off on scaling your AI workflows due to API costs, this might be the tipping point you've been waiting for.

Have you tested the new Sonnet yet? How does it hold up against your current workflows? Let’s discuss in the comments! 👇

03/07/2026

Claude Fable 5 just came back online. Most people will waste it on easy tasks.

Here is how to actually get value out of a frontier model while it is still accessible without per-use pricing.

1. Give it the problem you have been avoiding, not the one you already know how to solve. Fable 5 is built for the kind of work that takes a human days, not minutes.

2. Send one comprehensive brief instead of twenty fragmented prompts. Full context, full constraints, in the first message. Frontier models reward depth, not drip-feeding.

3. Default to a high reasoning setting. Save the maximum setting for the handful of problems that genuinely need it.

4. Give it a persistent memory file. A simple markdown log where it records what worked and what did not between sessions turns every future run into a slightly smarter one.

5. Instruct it to verify its own output before reporting progress. This single rule eliminates most false "done" signals.

6. Retire prompt structures you built for older models. What worked to control a smaller, more literal model often constrains a frontier one instead of helping it.

7. Use Fable 5 to write the operating instructions for the models you use every day. Let your best available model design the playbook, then hand that playbook to the model you actually run at scale.

That is leverage most people are leaving on the table.

The pattern underneath all of this: treat frontier access as a limited window to upgrade your systems, not just to finish today's task list.

If you are building AI-powered growth systems, this is the kind of operational thinking that compounds.

03/07/2026

I watched someone burn through their weekly usage limit in twenty minutes.

They were using Fable 5, Anthropic's most expensive model, to sort customer support emails into three categories.

That is like hiring a surgeon to put on a band aid.
Claude now ships in four tiers, and most people default to whatever model was last selected instead of picking on purpose.

Here is the quick version:

Haiku 4.5 is built for speed and volume. Classification, tagging, short replies, anything you run at scale.

Sonnet 5 is the daily driver. Writing, coding, analysis, research. Start here if you are not sure.

Opus 4.8 is for hard problems. Complex code, dense analysis, multi step reasoning where a wrong answer is expensive.

Fable 5 is the frontier tier. Long, ambiguous, high stakes work where the model plans its own steps with fewer check ins.

There is also an Effort setting on every model, from
Low to Max, that controls how hard it thinks independent of which tier you pick.

The chart below shows how the four tiers compare on capability, speed, and API pricing.

Matching the model to the task is the single easiest way to get better output and stop wasting your limit on work that never needed it.

What is your default model, and has it changed in the last few months?

02/07/2026

Claude Fable 5 is back. Here is what it means if you actually use Claude for work.

On July 1, Anthropic restored global access to Fable 5 after an 18 day suspension triggered by a U.S. export control order.

The short version: a jailbreak report showed the model could be prompted to identify software vulnerabilities and, in one case, produce exploit code.

Anthropic's testing found weaker models could do the same thing, but the government still required the pullback. Anthropic responded by training a new safety classifier that targets that specific technique, and the controls were lifted on June 30.

What matters for the average user is simpler than the regulatory story.

For most work, writing, research, planning, and general coding, nothing changes. Fable 5 runs as intended and handles long, multi-stage tasks better than prior models, sustaining context and checking its own output across complex work.

The one place you might notice a difference: security-adjacent debugging or prompts that resemble the reported jailbreak pattern. Those now get quietly rerouted to Opus 4.8, with no extra charge.

If you are on Pro, Max, Team, or select Enterprise plans, Fable 5 counts toward up to 50 percent of weekly usage limits through July 7. After that, it moves to usage credits.

Worth trying before that window closes, especially if you have been putting off longer or messier projects because previous models could not sustain them.

30/06/2026

DeepSeek V4 Flash costs $0.28 per million output tokens.

Claude Opus 4.8 costs $25.

That's an 89x gap, for models that are within single digits of each other on most coding benchmarks.

I broke down what this means for how companies will actually buy AI through the rest of 2026.

Curious where others are already routing between model tiers versus paying premium pricing across the board.

29/06/2026

Claude just posted a 75% increase in paying consumers since January 2026.

For context: ChatGPT has dominated this market for years. The shift happening right now is not incremental.

Credit card transaction data from Indagari, covering 28 million U.S. consumers, shows Claude's paid user base growing steadily month over month. On DataCamp, a platform with 20 million learners, "Claude" is now the most searched term on the site. More than "AI."

Among self-directed learners, demand for Claude courses is outpacing ChatGPT by 3 to 1.

Here is what this actually means for startups and growth teams.

The people choosing Claude are not casual users. They are builders, founders, and knowledge workers who use AI as a core operating layer. They are the exact audience making software purchase decisions, hiring fractional operators, and selecting the tools their companies will scale on over the next 12 to 18 months.

When the learner market shifts, the enterprise market follows. Usually within a year.

If your startup is still positioning against "AI users" as a monolithic category, this data is your signal to go narrower. The Claude user is a distinct persona with distinct expectations around depth, reasoning quality, and workflow integration.

That persona is now the majority among self-directed paid AI consumers.

The platforms, tools, and services that align with how Claude users work will have a structural advantage in the next growth cycle.

Are you building for them?

26/06/2026

We assumed AI would save time.

My research suggests something more interesting is happening.

AI does not simply free up hours. It redistributes cognitive effort - shifting knowledge workers away from output generation and toward quality assurance and strategic judgment.

That is a fundamentally different kind of work. And most organisations are not prepared for it.

I surveyed 150 knowledge workers to understand how AI adoption is actually changing time allocation. The findings are now published as a working paper on SSRN.

Key takeaway: the bottleneck is no longer production. It is judgment. And judgment cannot be automated - it has to be developed.

If you are leading a team, building a product, or advising a business through AI adoption, this is worth 10 minutes of your time.

I would be glad to hear how this maps to what you are seeing in your own work.

The link to the research paper in the first comment.

Address

Website

https://www.dominicbanguis.com/

Alerts

Be the first to know and let us send you an email when Dominic Banguis, MSc - Marketing & AI Growth Systems posts news and promotions. Your email address will not be used for any other purpose, and you can unsubscribe at any time.

Contact The Business

Send a message to Dominic Banguis, MSc - Marketing & AI Growth Systems:

Want your business to be the top-listed Advertising & Marketing Company?

Country:

City: