Google Flow 2.0 adds an agent, consistent Characters and buildable Tools that speed up AI filmmaking.
Google added a new AI agent that brainstorms ideas with you, handles multi-step tasks, batch edits an entire project and creates very long videos. It also added Flow Tools, which let you vibe code your own custom creative tools without writing a line of code, and it solved consistent characters.
Clicking the agent option opens a chatbot designed for content creation. Asked for three images of a tree in 3D, 2D anime and clay animation styles, it analysed the prompt, wrote detailed optimised prompts behind the scenes, and generated all three inside a side panel.
Not unless you let it. Before generating it asks permission: the three image batch was quoted at zero credits, a four second video at 15 credits. A toggle gives the agent free reign to generate without asking, but he keeps it off because it can burn through credits.
Gemini Omni on its own is currently limited to videos about ten seconds long. Combining the agent with Gemini Omni or the Veo 3.1 model gives you much longer explainer videos: the agent reads and understands your script, creates the different scenes, and stitches them together for you.
On its own the AI struggles to hold a character from scene to scene, so Google added a feature called Characters. You design a custom character that stays perfectly consistent across your videos and integrate it with the agent, either uploading an existing picture or generating a brand new one.
Click the select voice button on the character. You can pick from a wide variety of default voices or build a custom one: he named his Medieval Soldier and described a male around 25 with a deep British accent. You must still pick a base voice, here Karen, then save.
After clicking done, click the three little dots on the character and select add to prompt. Once he was linked, the agent was asked for three images of that character from different angles plus a fourth riding a horse, and the facial features and armour stayed completely consistent.
The Tools tab holds a community library of custom tools people built for AI filmmaking and content creation, including Simple Sketch, the image editor, Shot Explorer, Mask Magic, Type Overlays and Shadow Fix. A create tool button lets you design your own utility and publish it to the community.
It gives advanced control: a blur slider, saturation for colour grading, plus contrast and brightness. You can add a text layer with adjustable line and letter spacing and the default Google Sans font. Enable Depth analyses the image in 3D so the depth offset tucks text behind the foreground object.
Click create tool and describe it in plain language. He asked for a tool that converts any uploaded image into clay animation style, and the model built it with a Generate Claymation button that uses Nano Banana Pro. The output was not flawless, but the backend prompt can be tweaked.
“So Google just dropped a massive update to Google Flow, and this is basically Flow 2.0.”
“Personally, I don't like leaving that auto-approve button on.”
“Sometimes the agent can go a bit wild, generate way too much content, and quickly burn through a lot of your credits.”
“Gemini Omni is currently limited to making videos that are only about 10 seconds long.”
“Being able to dynamically tuck text behind an object without opening up Photoshop or After Effects is a massive time saver.”
“I can confidently say that these tools will absolutely revolutionize my personal filmmaking workflow, allowing me to produce high-quality AI content so much faster.”
“The New Google Flow Just Changed AI Filmmaking Forever.” https://www.youtube.com/watch?v=5AquJxcBGao. Transcript summary by WhipScribe, https://whipscribe.com/transcript-pages/the-new-google-flow-just-changed-ai-filmmaking-forever.html.