As with the last mid-year update, this Hardcore Data update also falls on July 4th. Even though Hypermatic is based in Australia, and not the United States (or Japan, as some seem to think, based on the design of this website), I do like the idea of doing these mid-year updates on Independence Day, and re-focusing it on continuing to declare our independence from legacy design/development workflows.
Plugin update highlights in 2026 (so far)
There are multiple updates and small fixes shipped to multiple Figma plugins on a daily basis, but here are some cool highlights from the first half of 2026:
- Added 120 pre-built email layouts to Emailify
- Added Canva exports to Convertify
- Added Adobe InDesign exports to Convertify
- Added Adobe InDesign imports to Convertify
- Added Customer.io API integration to Emailify
- Added Lottie file imports to Convertify
- Added Live GIF Timeline Previews to Bannerify
- Added Split Text Animations to Bannerify
- Added Column Background Images to Emailify
- Added CSV Variant Exports to Bannerify
- Added Preview Wall Mode to Bannerify
- Added AMP Email support to Emailify
- Added Synced Timeline Animation Previews to Bannerify
- Added Automated ChatGPT Translations to Emailify
- Added Selective Slide Loading to Pitchdeck
- Added custom HTML preview window resizing to Emailify
- Added Adobe Photoshop PSD exports to Convertify
- Added Scrollable Text Layers to Bannerify
- Added Bulk Edit Mode for Layer Settings to Emailify
- Added a Persistent Settings Panel to Emailify
- Added AI Prompt Animation Generator to Bannerify
- Added Real-Time Live HTML Previews to Emailify
- Added Adobe Photoshop PSD imports to Convertify
- Added HTML generation performance upgrades to Emailify
- Added Live Layer Animation Previews to Bannerify
- Added Multi-Scene Banner Support to Bannerify
- Added Grouped Animation Sliders to Bannerify
- Added clickable links for PDF exports to Emailify
- Added Animated GIF Previews to Emailify
- Added 500+ new email components to Emailify
- Added PPTX Shapes Support to Pitchdeck
- Added Airtable text export/re-import support to CopyDoc
- Added 20+ new email templates to Emailify
- Added Zapier integration to Emailify
- Added Zapier integration to Bannerify
AI Ops Melbourne Meetup
Earlier this year, in February 2026, I started the AI Ops - Melbourne Meetup to bring together designers, developers, founders, marketers and other people in Melbourne who are optimistic about AI, and who are already using it to automate real work.
The meetup has been a useful forcing function for thinking more clearly about what AI is actually changing in small companies like Hypermatic.
The most obvious answer is “tools”, but I don’t think that is the right level of abstraction.
AI Ops is not about tools. It is about workflows.
AI Ops means turning repeatable business processes into reliable AI agent workflows, supported by clear context, rules, tools, judgment and feedback loops.
The model is important, but the model is not the whole system. The system is everything around it: the instructions, the examples, the files, the tools, the review process, the way tasks get broken down, and the way the human operator decides what is good enough to ship.
AI multiplies the company you already have.
AI does not magically fix broken systems, bad culture, unclear priorities, or bureaucracy. It tends to amplify whatever is already there.
A simple, self-serve, low-friction company gets more leverage from AI because there are fewer layers for the work to pass through. A company that is already drowning in meetings, approvals and process will probably use AI to generate more meetings, approvals and process.
That has been one of the biggest lessons for me this year. AI is not just a productivity upgrade. It is also an organizational mirror.
Context is infrastructure.
Files like AGENT.md, DESIGN.md, reusable skills, product notes, support examples and internal checklists are becoming the new operating manuals.
They teach agents how the business works, how the product should feel, what trade-offs matter, and how repeatable tasks should be done.
In the past, a lot of this context lived implicitly in someone’s head. Now, the more of it you can make explicit, the more useful your agents become.
Workflows become assets.
As models improve, the valuable thing is not just the model itself. It is the repeatable workflow you have built around it.
Customer support, bug fixes, docs, billing, compliance, marketing, research, release notes, internal tooling - these are all workflows that can be gradually improved, systematized and reused.
The workflow is the asset. The model is the engine that keeps getting better underneath it.
The operator still owns the judgment.
Agents can investigate, write code, summarize logs, draft emails, update docs and propose fixes. But the human still verifies, deploys and takes responsibility.
That part has not gone away. If anything, it matters more now, because one person can suddenly move much faster and make much bigger changes.
The simplest practical rule I keep coming back to is this: if you do something more than twice, consider turning it into a reusable skill, checklist, prompt, script or agent workflow.
Figma plugins in the age of generative AI
A couple of weeks ago, at Config 2026, Figma announced a new native feature called “Generative Plugins”, which lets users prompt the Figma agent to build custom plugins inside their design files.
The basic idea is that you no longer need to understand code or technical terminology. You describe what you want, and the Figma agent builds a plugin for you.
One of the examples in Figma’s documentation is a prompt to build a complete accessibility checker plugin. That is interesting, because accessibility checking is already something many existing Figma plugins from the community can do, including some paid plugins.
So the obvious question for anyone making Figma plugins is: what happens when users can generate their own?
I think the answer depends heavily on the type of task the plugin is automating, and how much complexity is hiding underneath the surface.
Some plugins make perfect sense to generate. Anything with a low surface area of inputs and outputs is a good candidate.
Modifying Figma layers
- Rename layers by editing the layer
nameproperty - Tidy layers by editing the layer
xandyposition properties - Apply a small set of rules across a selected group of layers
Creating Figma layers
- Generate a soundwave visualiser
- Create mockups of designs inside device frames
- Draw a simple set of placeholder components
For that kind of task, generative plugins are a great fit. They give designers a faster way to create small, custom utilities for their own files.
But there is still no such thing as a free lunch. Everything has trade-offs, and everything has a cost.
For tasks that are not a simple one-to-one input/output transformation, the complexity grows quickly. The number of edge cases grows with it. The plugin no longer just needs to do one thing once. It needs to do the right thing repeatedly, across messy real-world files, weird layer structures, different team conventions, changing platform APIs and export formats that all have their own quirks.
That might be fine for teams with a dedicated AI Ops person or someone technical who enjoys debugging generated tools. But realistically, most designers, developers and marketers using Figma are already a day behind their deadline, stuck in meetings, and trying to maximize the small amount of actual work time they get each week.
They do not necessarily want to become maintainers of a custom-generated plugin. They want to get the work exported, delivered and approved.
What this means for Hypermatic
I have no delusions that AI agents, generative plugins and the broader shift toward more custom software will have no impact on Hypermatic.
I am sure we have already lost customers for certain plugins because they can now vibe-code a smaller version that fits their specific use case. Especially if they only needed a small subset of a plugin’s functionality in the first place, and that’s okay.
It is also not entirely new. Hypermatic has 12 different Figma plugins, each tackling a different use case. Over the years, plenty of plugin features have eventually become native to Figma itself, including password protection, spell check, Figma Slides, Figma Dev Mode and many more.
Whether a feature becomes native to Figma, gets replaced by a generated plugin, or gets rebuilt internally by a team using AI, the outcome is similar: the lowest-complexity parts of the software market become more abundant and less valuable over time. That has always been true. AI just makes it happen faster. The important question is not “can someone generate a plugin that does this?” The better question is: “do they want to own everything that comes after generating it?”
Software as a Service
When people talk about SaaS, they usually focus on the software part. That is understandable, especially with all the headlines about the impending “SaaSpocalypse”. The basic argument is that if anyone can vibe-code software with AI agents, the value of software companies falls off a cliff, because every customer will simply clone all the SaaS products they currently pay for and bring them in-house.
That will be true in some cases. There will be plenty of internal tools, dashboards, workflows, scripts and small utilities that no longer need to be bought from an external vendor. If the thing is simple enough, specific enough and low-risk enough, it probably should be generated internally.
But I think the bigger argument throws out the baby with the bathwater. In many cases, you are not just paying for the software itself. You are paying for everything that comes with the software.
You are paying for:
- A dedicated company and team behind the product
- Customer support when something breaks or needs explaining
- Years of edge cases that someone else has already found and fixed
- Documentation, examples and onboarding material
- Ongoing maintenance when APIs, platforms and file formats change
- Infrastructure, hosting, security and uptime
- Someone else to care about the boring details forever
A lot of software is valuable precisely because you do not have to think about it anymore.
Things I would still rather buy
If we flip this around, there are plenty of things I would not want to create, own, maintain or run as vibe-coded versions myself.
For example:
- Payment rails or anything to do with finance
- Servers, hosting and infrastructure
- SMTP servers or anything to do with sending and receiving email
- Real-time messaging apps like Slack or WhatsApp
- Design tools like Figma and Canva
- LLMs or code harnesses like OpenAI or Anthropic products
- Video editing and rendering tools
- Anything that processes or stores sensitive data
All of these products have many years, and in some cases decades, of thought, design, robustness and edge case handling built into them.
The idea that I would choose to vibe-code my own version of any of these from scratch, then run a critical part of my business on it, just to save $100/month while burning thousands of dollars in tokens and future maintenance time, feels like the wrong trade-off. AI makes it easier to build software. It does not make all software free to own.
Where I think this lands
AI agents are leverage. In the same way that you could always code your own internal tools before, you can now spin up the first version of those tools much faster. That is genuinely useful, and it will change which products people choose to buy. But the first version was never the hard part of software. The hard parts are everything after that: handling edge cases, supporting customers, maintaining compatibility, improving the workflow, updating the product as platforms change, and making sure it keeps working when people rely on it.
That is where I think Hypermatic still fits. The easy parts of software are becoming cheaper every month. The difficult parts - understanding messy real-world workflows, handling accumulated edge cases, earning trust, and continually improving products over many years - are becoming more important. That is the direction we are continuing to build toward. Smaller company. Better workflows. More leverage. Less backlog.
Pro Bundle