In partnership with
Have you tried Google Flow yet?
Google Flow has generated 275 million videos since its launch in May 2025, and yet you probably haven't heard of it.
I hadn't either until recently. And when I did, I made time to test it.
I created a noir detective sequence from scratch without a camera crew or editing experience.
Here's what happened. But before that let’s catchup on AI this week:
News News News:
TOOLS that caught my attention
1. Memos 2.0
Open-source memory operating system for AI agents that stores, retrieves, and manages conversational context across sessions.
2. Medeo AI
AI medical scribe that listens to doctor-patient conversations and auto-generates clinical notes in real-time.
3. SkillStack
A curated, paid marketplace for Claude Code skills, agents, and workflows where creators set prices and every upload is verified before going live.
What flow is
Flow is Google's AI filmmaking tool, built on top of Veo 3.1.
Think of it as directing camera shots through text prompts.
You can generate and control scenes, camera movements, and visual style.
It’s uniqueness lies in a feature called Ingredients.
You can upload reference images of your character, and Flow maintains its visual consistency across multiple scenes.
The big caveat is its pricing: $19.99/month gets you 100 generations. And serious use requires you to upgrade to the $249.99/month Ultra tier.
Building a noir sequence
I wanted to see if Flow could handle cinematic storytelling.
I chose the film ‘noir’ because its lighting and atmosphere matter more than complex interactions.
Scene 1: The Detective at His Desk
Started with "Frames to Video" mode. Uploaded a single character reference: 1940s detective in fedora.
My prompt structure:
Subject: "A lone detective"
Environment: "in a noir office"
Action: "examining case files"
Camera: "medium shot, slow push-in"
Mood: "Dramatic lighting from desk lamp, rain visible through window"
Result:
Flow nailed the atmosphere.
The interplay of light and shadow is excellent. (blue rain tones contrasted with warm desk lamp and pink neon sign outside)
Character stayed recognizable and the face didn't morph. And the textures like the wrinkles in the shirt held up.
The illusion breaks:
Watch his hands interact with paper here. The paper doesn't behave like paper, instead it bends like rubber.
This is the current state of AI video: it understands what things look like, but not how they behave.
Scene 2: Close-Up of Hands (Three Attempts)
First attempt: weird finger positioning.
Second attempt: closer, but folder angle was off.
Third attempt: nailed it.
I regenerated the same prompt three times.
Flow gave me three different interpretations.
AI doesn't know which version is better, hence the directing skill becomes important here.
Scene 3: Wide Establishing Shot
Prompt: "Wide shot of noir detective office showing full desk with neon signs visible outside rain-streaked window, static camera, moody blue and pink lighting, establishing shot."
Got it first try.
Although the transition…
Need I say more?
Overall, the character consistency held up. And this is Flow's biggest win over other tools.
Final Stats:
Total time: 28 minutes
Scenes generated: 12 (used 4 in final sequence)
Credits used: 12% of monthly Pro plan allowance
Cost: $19.99 (Pro tier)
Who this actually helps
Flow is perfect for:
Concept films (show investors your vision in 30 minutes instead of 3 weeks)
Marketing visuals (test 5 style variations in an hour)
Rapid prototyping (validate ideas before expensive production)
Branded content with recurring characters
Flow is terrible for:
Feature-length anything (credit limits kill this)
Dialogue-heavy performances (emotional realism isn't there)
Long-form shots over 4 seconds (you'll stitch constantly)
Complex character interactions (physics break down)
What worked (and what didn’t)
My take
A cinematic sequence that would've taken weeks and thousands in equipment to shoot traditionally now just took me 30 minutes.
Stock footage sites selling office shots at $50/clip are competing with AI at $0.20/generation now.
But Flow still doesn't know when to break rules for emotional impact.
I regenerated my close-up three times before realizing AI picked the wrong one every time.
That's still a solvable problem though.
Google’s graveyard problem isn't.
They have killed 299 products till date.
Recent casualties: Chromecast, Google Podcasts, Jamboard. And VideoFX i.e Flow's direct predecessor was already discontinued long back.
This is a pattern: launch ambitious tool, generate excitement, fail to commit, kill it 18 months later.
The naming confusion too (Workspace Flows, Flow AI, various "Flow" products) is a symptom of Google not treating experiments seriously.
Flow's tech is impressive. But that’s still low stakes. We’re betting that Google won't pull the plug on it.
So until they do: Flow is the best tool for consistent-character stylized content. Give it a try. And reply with the link to your output.
Until next time,
Vaibhav 🤝🏻
If you read till here, you might find this interesting
#AD 1
Better prompts. Better AI output.
AI gets smarter when your input is complete. Wispr Flow helps you think out loud and capture full context by voice, then turns that speech into a clean, structured prompt you can paste into ChatGPT, Claude, or any assistant. No more chopping up thoughts into typed paragraphs. Preserve constraints, examples, edge cases, and tone by speaking them once. The result is faster iteration, more precise outputs, and less time re-prompting. Try Wispr Flow for AI or see a 30-second demo.
#AD 2
AI-native CRM
“When I first opened Attio, I instantly got the feeling this was the next generation of CRM.”
— Margaret Shen, Head of GTM at Modal
Attio is the AI-native CRM for modern teams. With automatic enrichment, call intelligence, AI agents, flexible workflows and more, Attio works for any business and only takes minutes to set up.
Join industry leaders like Granola, Taskrabbit, Flatfile and more.
