The Syntic Charter

Every company is shaped by what it refuses to compromise on. This document is the list of things we won't compromise on at Syntic — written down so customers, employees, partners, regulators, and our future selves can hold us to them.

Syntic is an independent AI company. We've built our own frontier model, Amara, from the floor up. We sell two products: a complete consumer AI assistant and a platform for hiring AI Employees. We don't take outside investment. We don't have a board appointed by venture capital. We answer to the customers paying us and to the integrity of what we're building.

This Charter is not marketing copy. It is the working document we refer to when product, business, and ethical decisions come into conflict. When you read about a Syntic choice that seems odd or non-obvious — a feature we won't ship, a customer we won't take, a price we won't raise — the reasoning will be in here.

It is a living document. We will revise it as we learn, as conditions change, and as we get things wrong. Past versions remain on the public record. We don't quietly edit our principles.

Syntic is three things at once. Understanding which one you're talking about makes the rest of this document make sense.

Syntic AI, the company

An independent, self-funded technology company. Twenty-six people across the world. No outside investors. No board imposed by capital allocators. We are accountable to our customers, our team, and the law of the jurisdictions we operate in. We are not accountable to an exit timeline.

This structure is a feature, not an accident. We designed the company to be able to make multi-decade decisions. Most of the AI industry can't.

Syntic, the platform

The software and infrastructure we build and operate. Two products today — the Syntic assistant and Syntic Workforce — plus the platform underneath that makes both possible: the model, the runtime, the safety layer, the integrations, the deployment surface.

The platform belongs to the company. The work done on it belongs to the customer. That distinction matters and runs through everything else in this Charter.

Amara, the model

The frontier AI model family we trained from scratch. Amara is the engine inside everything Syntic ships. We use the name "Amara" for the model because the model is its own thing — distinct from the products that wrap it and the company that built it.

When we talk about Amara's behavior, we're talking about the model. When we talk about Syntic's behavior, we're talking about the company and its products. Sometimes the lines blur. We try to be precise about which one is doing what.

We serve four kinds of customers and we owe each of them a different set of commitments.

If you don't see yourself in this list, we can probably still help you — but the four groups above shape how we make decisions.

The convictions that shape every product, pricing, and partnership decision at Syntic. Each one is something we'd defend in front of a customer, a journalist, or a hostile cross-examination.

Customer work belongs to the customer

The AI workforce inside a customer's organization is theirs. Their data. Their custom models. Their AI Employees. Their conversations. Their voice infrastructure. Their workflows. We are the platform; we are not the owner of the work done on it.

This is contractual, not aspirational. We don't train shared models on customer content. We don't sell customer data. We don't permit advertisers to influence Syntic outputs. We don't keep customer data after the customer asks us to delete it. We don't claim ownership of derivative work produced through our platform.

If the company is ever acquired, sold, restructured, or wound down, customer ownership of their work survives the transition. This is written into our terms of service and it's a line we will defend in any negotiation.

Independence is a feature

We don't answer to venture capital. We have no board appointed by capital allocators. We are not for sale. Any capital that enters Syntic does so on terms that preserve those facts — or it doesn't enter.

This isn't a slogan. It's the structural property that lets us make every other commitment in this document credibly. A company accountable to venture investors has to satisfy a return profile that's measured in quarters. A company accountable to its customers has to make products customers actually want to keep paying for.

We have customers in 40+ countries, on six continents, paying us for products we built with our own money. That arrangement keeps us honest in ways no governance structure can replicate.

If a strategic relationship with outside capital ever makes sense, this document is where we'll explain why it does and what it protects — before we enter it, not after.

Own your core technology

The companies that matter long-term will be the ones that own the technology they're selling. We invested years in building Amara from scratch — pretraining, alignment, evaluation, infrastructure, voice, safety layers — because the alternative was to be a product company built on a competitor's research.

That alternative doesn't survive a single API price increase, a policy change, or a competitive acquisition. We refused to build a business on someone else's foundation. The model is ours. The infrastructure that serves it is ours. The safety layers around it are ours.

This belief has costs. Building a frontier model is expensive and slow. We absorbed those costs because they're cheaper than the long-term cost of dependence.

Honesty is operationally critical, not just morally correct

AI companies are in a category-defining trust position with their customers. The asymmetry between what we know about our systems and what customers can verify is enormous. Honest accounting of capabilities, limitations, certifications, incidents, and intentions is how we earn the trust the asymmetry requires.

What this means in practice:

• We don't claim certifications we haven't earned. SOC 2 Type II is "in progress" on our site because it is. We will say "certified" the day after the certificate is issued, not before.
• We don't fabricate customer counts, revenue figures, or growth metrics. We don't publish customer logos without permission. We don't quote testimonials we wrote ourselves.
• We don't ship features that look better in a demo than they perform in production. If a capability is a roadmap item, we say so.
• We publish incidents. When something material goes wrong, customers learn about it from us first, with honest accounting of what happened, what was affected, and what we're doing about it.
• We don't manipulate our own search results, public messaging, or analytics to project a different version of the company than the one we are.

This is the standard we hold ourselves to. We'll fail to meet it sometimes. When we do, the response is to correct the record publicly, not to bury it.

Humans stay in the loop on the decisions that matter

Syntic Workforce deploys AI Employees that handle real-world consequence — phone calls to customers, payments, contracts, code that ships to production, communications with patients, regulatory documents. AI judgment alone is not sufficient for these actions.

Our platform makes human oversight a primitive, not an afterthought. Approval gates on irreversible actions. One-click human takeover on every voice call and text conversation. Audit trails on every Agent Execution. Versioning and rollback on every AI Employee. Live monitoring on every active conversation.

These aren't features customers can opt into. They're properties of the platform. The customer chooses where to set the thresholds — we make sure the controls exist.

Compliance is a commitment to the customer, not a marketing badge

Customers in regulated industries trust us with workflows that have legal consequence. Collections calls subject to FDCPA. Patient communications under HIPAA. Citizen services under jurisdiction-specific privacy laws. Financial communications under model risk management frameworks. Voice campaigns subject to TCPA and equivalent national laws.

The relevant compliance frameworks are built into the products that touch them — not bolted on afterward. Where we don't yet have a certification, we say so. Where we have controls that align with a framework without formal certification, we explain the distinction. Where a customer needs a compliance posture we don't yet support, we tell them honestly and we don't sign the deal pretending otherwise.

If a regulatory regime in a jurisdiction we operate in requires us to change how Syntic works, we'll do it — and we'll publish the change rather than hide it inside product updates.

Price discipline is a form of trust

The Syntic assistant is $20 a month. We launched at this price. We intend to keep it at this price even as the industry around us raises subscription prices and adds tiers. We don't have venture-backed quarterly growth pressure to satisfy. We can hold prices.

For Syntic Workforce, our pricing scales with what customers put into production — not with their company size, not with seats they're not using, not with feature gates designed to extract more revenue from organizations that already pay us. Enterprise contracts include negotiated terms; our list pricing is honest.

When we do raise prices, we'll explain why — and we'll grandfather existing customers where we reasonably can.

Every company has lines it won't cross. The honest ones write them down.

The list below is not exhaustive — judgment will always cover cases we haven't anticipated — but these are absolute. They apply regardless of customer size, contract value, or commercial pressure.

If you're a customer or partner asking us to do one of the things in this list, we'll decline. We'll do it politely. We won't do it.

The operating practices that make the rest of this document possible.

Small team, large surface

Twenty-six people are running a frontier model program, two product surfaces, and customers in 40+ countries. This works because we're disciplined about what we build, who we hire, and how we operate. We don't grow headcount to look impressive. We grow it when there's work that genuinely requires more humans.

This means some things ship slower than they would at a 200-person company. It also means everything we ship is built by people who care about the outcome. The trade is worth it.

Distributed by design

Syntic has no headquarters. The team is across the world by intention, because the best people don't live in one city. This shapes how we operate. We default to written communication. We document decisions. We respect time zones. We don't romanticize being in the office because there is no office.

We dogfood our own product

The work inside Syntic that doesn't require human judgment is done by AI Employees on the Syntic platform. We use Forge to build them. We use Cowork to monitor them. We use the Call Center when we run customer outbound. We use Syntic Code to ship Syntic. This isn't a marketing line — it's the practical reason we can run a frontier AI company with 26 humans.

It also means we feel our own product's bugs first. When something is broken for our customers, it's usually been broken for us for a few hours already.

Ship what we publish

Our research team publishes openly when we can — on alignment, evaluation, multi-agent coordination, voice, autonomy thresholds. We ship what we publish: research findings become product features, evaluation harnesses become deployment gates, safety properties become runtime guarantees. The lab isn't a marketing function. It's how the platform improves.

Where customer security or competitive considerations require us to keep work internal, we keep it internal. The default leans toward open.

Syntic operates a frontier AI model, built from scratch, that powers everything described in this Charter. How that model behaves, what it values, what it refuses to do, and what we owe it as an entity — these are questions we take seriously enough to deserve their own document, not a section here.

The full account is in our Model Specification, the foundational document we wrote for the model itself and for anyone who wants to understand how it was built to think.

Read the Model Specification →

This Charter is a living document. It will be revised. Here's how that works.

How we amend

Material changes to this Charter are reviewed by the founder, the research lead, and the legal lead before publication. Changes are announced publicly when they ship, with a brief note explaining what changed and why. Past versions remain on the public record at syntic.ai/charter/archive.

We don't quietly edit our principles. If a future version of Syntic walks back a commitment in this document, we'll explain why before we do it — not hide it after.

How we're held to it

The accountability mechanisms for this Charter are:

• Customers can hold us to it through their contracts. Where this Charter is more protective of the customer than the contract, the Charter wins.
• Regulators can hold us to public statements about our practices. This document is a public statement.
• The press can hold us to it by checking. We invite scrutiny.
• Our team can hold us to it internally, including by leaving and saying so publicly if we materially violate it.
• The founder is named on this document and personally accountable for it.

If we fail to live up to something in this Charter, the response we owe is: acknowledge the failure publicly, explain what went wrong, fix it, and update the Charter if our understanding has changed.

Most AI companies in this cycle are wrappers — products built on borrowed technology, financed by borrowed money, accountable to investors who will pressure a sale within five years. They are racing each other to the same exit.

We're trying to build something else. An independent company, on a model we own, with products we control, serving customers we answer to. The Charter above is the operating manual for that attempt.

We won't get every line right. We'll learn things that make us update parts of this document. We'll occasionally fail to live up to commitments we made in good faith. When that happens, the Charter is what we'll come back to — both to remember what we said and to honestly reckon with the gap between what we said and what we did.

This Charter is the company explaining itself to itself. We publish it because customers, employees, partners, and our future selves deserve to see the explanation.