Prompts
Single-instruction templates that produce a specific output: a draft, an analysis, a reformat. Drop in, fill the variables, run.
Independent creators publish prompts, personas, skills, assistants, and agents. We test every listing before it goes live — so the person at the keyboard spends time using AI, not debugging it.
What Sigrix sells
A taxonomy that follows complexity. Pick the smallest unit that solves your problem — most people need a prompt, not an agent. Mix and match as your workflow grows.
Single-instruction templates that produce a specific output: a draft, an analysis, a reformat. Drop in, fill the variables, run.
System prompts that give a model a stable role, tone, and worldview — so a generic chat surface acts like a senior editor or a tax advisor.
Composable mini-capabilities you bolt onto an existing assistant — a calculator, a citation checker, a tone enforcer — to extend what it can do.
Persona + skills + memory, packaged as a working tool with a defined surface area. Hire one for a job; it stays in scope.
Multi-step systems that plan, call tools, and keep going until a goal is met. The most powerful tier — and the one that needs the most scrutiny.
Our mission
AI can only go as far as the people using it understand and guide it. Sigrix exists to close that gap — verified components, plain-English documentation, and discovery that matches the buyer instead of the highest bidder.
From the creator
My day job is data engineering in financial-services risk. I started out as a practicum student in IT, moved through IT support, and spent the better part of a decade in banking technology — first as a project officer, then a climate-risk business analyst, and now building the data pipelines behind mortgage risk models. The through-line is the same: take something messy and undocumented and turn it into a system people can actually trust.
I started Sigrix because I kept watching people pay for AI tools they couldn't get to behave — copying prompts off forums, hitting walls with agents that quietly failed, and giving up on workflows that should have been five minutes of work. The technology is here. The curation isn't. So I'm applying the same instincts I use on risk data — provenance, reproducible tests, no black boxes — to a marketplace of AI building blocks I'd actually want to buy from, with creators kept in the loop instead of mined for IP.
Outside of code I'm an artist — I've had several solo exhibitions — and what I like most is making things from nothing. Sigrix is exactly that: built from scratch, curated by hand, one listing at a time.
Strategic advisor
I joined because I kept seeing the same pattern. Organisations invest in AI tools, hand them to their teams, and wonder why nothing changes. The technology works. The understanding doesn't.
Fifteen years working across startups, SMEs, and large multinational corporations in transformation and change management taught me that tools only deliver when people know how to direct them. Prompts are no different — the gap isn't access, it's guidance.
What Sigrix is building is the thing I couldn't find elsewhere: a marketplace where every listing is actually verified, a co-creation model that respects the people building it, and an education module that doesn't just hand you a prompt — it helps you understand why it works.
That combination is what brought me here. I'm responsible for the strategy, the communications, and making sure the platform grows in a way that stays true to what we set out to build.
How we curate
Curation is the product. A typical listing spends 1–2 days in review. Test results and reviewer notes ship with the listing, so buyers see the same evidence we did.
Prompts, personas, or assistants come in with examples, model targets, and a signed creator agreement.
We read every listing for clarity, scope, and honesty about what it can — and can't — do.
Each listing is run against a fixed test set on the declared model. Output is graded for accuracy and consistency.
We screen for prompt-injection traps, unsafe outputs, and copied work. Failed checks come back with feedback.
Approved listings publish with a verified badge, full provenance, and the test results visible to buyers.
Founding seller program
Sigrix opened on May 15, 2026. A limited number of founding seats are still open — claim yours and keep the perks for life.
Seats close when the founding cohort fills. Every perk is locked for life.
Digital ownership
Listings are licensed, not absorbed. Sigrix takes a commission on the sale — that's the entire commercial relationship. We don't claim a perpetual license to your prompts, train models on them, or sublicense them to anyone else.
Buyers receive a permanent, transferable license tied to their account. If Sigrix shuts down tomorrow, you still have the files you paid for. See how it works →
The mechanism behind that is published, not ours to keep. Postern is the open protocol an agent uses to check the person running it is entitled to — drafted in the open under Apache-2.0, with Sigrix as one implementation of it rather than the definition. Credentials never travel through it: an agent declares the environment variable names it needs, so "your keys stay on your machine" is checkable instead of promised.
Get started
Browse the marketplace for AI agents, prompts, and assistants as a buyer, publish your work as a creator, or apply to the founding seller program for permanent perks. Every direction makes the marketplace better for the next person.