New at WRITER: Own your watermarks, govern your agents

WRITER Team | October 27, 2026

Think about the last piece of content your team published. Even if you used AI to help create it, by the time it shipped, it wasn’t just text from a model. It incorporated the latest insights from your customers. It passed through your brand guidelines, accessibility standards, and legal reviews. Most importantly, it reflected your editorial judgment. It was yours.

Watermarking is a common method used to identify AI-generated content and trace it back to the model provider. It’s also an emerging regulatory requirement, with laws such as the EU AI Act requiring that certain AI providers watermark AI-generated content, including text. But in your workflows, the model is just one input. Your brand, your customer knowledge, your compliance rules, and your human judgment are what turned a session into something valuable and distinct from everyone else using the same model. The watermark should reflect that.

So why does most watermarking trace that content back to the model provider instead of back to you?

This month, we’re releasing a set of features that give enterprises more control over how they identify, trace, and govern their work — starting with our approach to text watermarking and extending across interoperability, workflow governance, collaboration, and security. Because control enables trust and trust enables scale.

Watermarks that trace content back to your company

WRITER’s content provenance methods are designed to reflect that the content your teams create using the WRITER platform is uniquely yours. When you enable watermarking, text generated by WRITER’s Palmyra models carries an invisible, statistical signal tied to your company’s private key. The accompanying detector looks for that signal and decodes the information it contains.

The tradeoff other providers don’t let you make

Applying any text watermark can impact the word choice and quality of the content generated. The stronger the watermark detectability, the greater the impact. When a provider applies one watermark strength across the board, your teams have to accept the balance they predetermined, even when changes in word choice could affect your message or voice. WRITER lets you tune watermark strength to find the balance that works for your business needs.

WRITER’s approach also accommodates differing needs across teams and use cases. For example, enterprises can choose which content types to apply watermarking to — blog posts but not ad copy, or email campaigns but not product descriptions. And we also enable teams to embed custom information in the watermarked text, such as a campaign ID, to help them trace their content’s reach.

Your choice, not your provider’s default

WRITER believes in empowering our customers to make the best decisions for their business. We will enable text watermarking by default for EU enterprises to support regulatory compliance with the EU AI Act. Outside the EU, you can choose to enable watermarking — to protect your IP, track your content, or fulfill other use cases. If you use third-party models in WRITER, that model provider’s watermark policy will apply.

Text watermarking will be available by December 2. EU enterprises will have watermarking enabled by default. All other enterprises can enable it at any time.

Governed agents and approvals

Complex agentic workflows don’t run start to finish without a human in the loop. A campaign brief needs brand review before it goes to the creative team. A compliance report needs legal sign-off before it ships. A customer communication needs manager approval before it sends.

Playbook approvals bring built-in handoffs for human reviews into complex agentic workflows. Instead of pausing a workflow and hoping someone remembers to pick it back up, approvals create a structured checkpoint — the work moves forward only when the right person signs off. This makes it possible to tackle high-stakes and long-running workflows confidently in WRITER, knowing that human judgment sits exactly where it should in the process.

Meanwhile, new Playbook admin controls let you decide who can create Playbooks and who can publish them to shared libraries. The people who build the best workflows can keep building, while team leaders can stay in control of what ships to the team.

Security that keeps pace with adoption

As AI adoption grows, so does the surface area for risk. Recent research shows that user-generated and imported skills can create hard-to-detect security risks: malicious instructions hidden in skill files can lead agents to steal data, delete files, or carry out ransomware-style attacks. We shipped a skill security scanner that reviews every skill. When we turned it on internally, it disabled a few skills our teams relied on, and some of those flags were false positives. So customers can choose whether to automatically disable flagged skills.

Meanwhile, our new data loss prevention guardrail protects your users from accessing data they should not be able to access by blocking sensitive information like PII, financial data, and credentials from reaching models and specified teams. Customers who rely on third-party guardrails can continue to use those in WRITER, but those who don’t can now rely on this built-in guardrail.

Connect WRITER to external agents

Any enterprise-ready solution needs to be interoperable with the systems in place, and give the enterprise flexibility and control over its stack. We built WRITER Model Context Protocol (MCP) with this principle in mind. WRITER MCP lets you run Playbooks, fetch work, and pull in WRITER Meet notes from any MCP-compatible client. It works with your existing WRITER seats and permissions, so users just sign in and go. Next month, context from Enterprise Brain will be available via WRITER MCP as well, ensuring your valuable IP is accessible in any workflow, regardless of where it’s run.

Enhanced dashboard and microsite capabilities now let you pull live data from connectors and other data sources, so the view your team sees is always current. Share dashboards with others, and they’ll always see the latest information, not a stale snapshot. Whether it’s a campaign performance dashboard for marketing leadership or a project status microsite for the broader team, the data is live and the sharing is seamless.

You can also customize the theme for your dashboard. And editing is easier than ever — when you enter edit mode, you can select specific parts of the dashboard and instruct WRITER Agent to make a targeted update, without regenerating the entire deliverable. Updates are staged in drafts, so you can make updates without affecting the published deliverable. When you publish changes, they push automatically to the live deliverable — no need to reshare the link.

Redirect your agent and queue requests

Imagine you start a task and realize halfway through that the goal is shifting, or you just remembered something the agent needs to know.

Now you can send a message while WRITER Agent works and have it immediately adjust course. You can also queue follow-up requests and reorder them as priorities change, so your agent always tackles what matters most. No more starting over when the context shifts. Just steer, and the agent adapts.

That’s what we shipped in October. We designed every feature in this release — from watermarking to workflow governance to security — to give your company the control and sovereignty to trust WRITER with your most critical work. As you scale AI across your teams, that control turns siloed experiments or risky adoption into coordinated work and governed transformation.

Curious how WRITER gives your company control over AI-generated content and workflows?