The Upgraded Hedra Agent: Built for Full Creative Workflows

For the last year, we have been building agents that help you create. The first version of the Hedra agent was a better interface to asset generation. You typed a request, it produced an image or a video, you iterated. It was useful.

But the way creative work actually happens is bigger than asset generation alone.

A solo founder planning a launch needs to analyze the market, write a script, generate visuals, produce a video, and post it. A creative director pitching a new idea needs competitive context, a creative direction, mood boards, and a film.

An educator creating a course needs source material, a structure, illustrations, voiceover, and a finished module.

The work spans research, reasoning, writing, generation, and delivery. Today, almost no one does that work in one place.

Today we are upgrading Hedra Agent.

It is now a general agent built for creative work. It can research, reason, and plan. It produces professional video, image, and audio. It delivers the finished output. All in one workflow and in one space.

The creative stack is broken

Most creatives we talk to are operating across a painful patchwork of tools:

• Research happens in ChatGPT or Claude

• Scripts and briefs get written in a doc

• Assets are produced in creative tools

Everything ends up in scattered docs, Notion pages, and Drive folders. For teams, none of it is shared or coordinated

The result is a deep separation between research and production.

Strategists cannot execute. Designers cannot be strategic. Solo operators are bottlenecked by their own bandwidth.

The most ambitious people we know have tried to fix it themselves by stitching together pipelines in tools like Make or Claude Code. That works, but it requires real technical chops, and it still involves a lot of manual handoffs.

The fundamental problem is that the workflow is fragmented across products that were never designed to talk to each other.

We have spent the last year asking what it would take to fix that.

Hedra Agent is our answer.

A general agent that ships finished media

Hedra Agent is closer in spirit to general agents like Manus or Hermes than to any existing creative tool. Its core capabilities include:

• Planning multi-step work and running concurrent tool calls

• Reading and writing across your connected systems through MCP

• Running code in a sandbox as part of a workflow

• Doing deep research on the web with tools like Exa

• Executing scheduled tasks autonomously on a cron

• Accessing the best media models available, with the intelligence to choose the right one for the job

The critical difference is that the research powers the production.

Other general agents handle research and strategy. Other creative tools handle asset production. Today, the work moves between them, with exports, handoffs, and scattered files in between.

The Hedra agent uses its deep research to create the assets. When it analyzes a competitor's brand, it generates visuals informed by what it found. When it pulls trends from Reddit, it scripts UGC that reflects them. The research feeds the work directly.

Video, image, and audio come out as first-class outputs, with the agent choosing the right model for each job.

We are the only general agent that uses its research to create finished media.

One agent. One workspace. Finished work.

Spaces: a persistent workspace, not a chat window

The primary surface in Hedra is a Space.

It is a persistent, living canvas the agent writes to and reads from as it works.

Inside a Space, the canvas holds text blocks, images, videos, and research. The agent treats them as part of its working context.

When you ask it to draft a script, it incorporates the brief. When you ask it to produce a video, it bases it on the script. When you ask it to generate visuals, it pulls from the research already on the canvas.

This is the value of Space. Different types of media live alongside each other, and the agent builds on them as the project develops. The canvas is the unit of work, not the chat.

Your research lives next to the brief. The brief lives next to the script. The script lives next to the video.

Everything is in one place, the way a project actually behaves.

Skills: your workflows, made reusable

Anyone who uses agents consistently ends up typing variations of the same request again and again.

Analyze the top performing ads from your Instagram last week. Update your slide deck with the latest product changes.

The repetition is natural. The hard part is remembering the exact wording and the order of steps that made the last result land.

Skills solve that.

A Skill is a saved playbook the agent follows whenever you invoke it. You give it a short name, a description, and a body of instructions written like a creative brief.

After that, you trigger it with a slash command and add only what changes for this run.

Skills are individual, but they are also shareable.

A team can build a library of Skills that encode how the team works. A creator can publish their best Skills for other creators to use.

The work you do once becomes leverage for everyone who comes after you.

We wrote a longer post on how Skills work and what they are good for. The short version is that Skills are how you turn one good output into a hundred.

What you can build today

The clearest way to explain Agent is to show what it can actually do.

Three examples a single user can run today:

The Space is the dashboard. You return to it whenever you want, and the context is still there.

The agent does the listening, the writing, and the production. You direct the strategy.

These are not aspirational. The agent can do all three today.

The hard part is no longer the tooling. The hard part is what you choose to build.

Built for the operator who needs to scale themselves

Hedra Agent is for the individual contributor who needs to operate at the scale of a team:

• Solo founders who have more vision than bandwidth

• Content creators producing high volumes of regular content

• Creative directors at agencies pitching new business

• Educators and journalists who need a research-to-publish workflow

• Designers who want strategic leverage

• Strategists who want to execute their own ideas

• Go-to-market teams looking to scale creative output without scaling headcount

The common thread is that they are looking for an edge.

They are early adopters. They are comfortable with AI. They are not necessarily developers, but they are curious and willing to learn.

They understand that the people who figure out how to use agents well right now will have a significant advantage.

What they want is not time savings. Time savings is the byproduct.

What they want is capability and creative range they did not have access to as a solo operator or a small team. They want to do things they literally could not do before.

That is who Hedra Agent is for.

If that sounds like you, this is yours.

Where this goes next

Today is the start, not the end.

The architecture we have built, with Spaces as the workspace, Skills as the automation layer, and the agent as the reasoning engine, is a foundation we will keep building on.

More integrations through MCP. Deeper research capabilities. A growing library of Skills shared by the community. Scheduled workflows that run while you sleep. The model intelligence to keep choosing the best media model for the job.

The platform gets better as the models get better. Your work gets better as your library of Skills grows.

The thinking and the making belong in the same place.

We are building the place.

Hedra makes it possible. What will you create?