We make AI better.
And actually easy to use.

Hand-tuned merls. Each one is ridiculously good at one thing so you don't have to be. You show up with the idea — they bring the taste.

Download for iOS → Android Use on Web
Finest AI models. ✦ Specialized for you.
The best AI models in the world — trained by us
✦ ✦

Specialists. One crew.
In your pocket.

Each agent is tuned by hand for one job. No generalists. No slop.

01 · SHINE 12 mana
Same face.
Better frame.
AI headshots. LinkedIn-ready. No uncanny.
"
No more AI slop. Just the thing you meant to make.
— the whole idea
02 · EDITOR5
Tailored
for the job.
Résumés + cover letters, mirrored to the posting.
03 · ECHO5
Your voice.
Not a robot's.
Tweets, threads, captions, essays, emails — in your cadence.
05 · DESIGNER12 mana
Same bones.
New soul.
Redesign any room. Keep the light.
04 · CUTTER30–60
Describe it.
We'll film it.
Best video models in the world. Trained by us — face/hand warps caught and re-rolled before you ever see them.
FREE FIRST CAST
Try every agent, free.
One cast per agent on the house. No trial clock, no credit card to start.
· Proof · agents · real outputs ·

Specialists read the room.
Generalists just finish the sentence.

Same task. Same inputs. Different taste. Tap through each.

Task · tailor a résumé for a Senior PM role at a Series B fintech
Generic AI ✗
Sarah Kim
Senior Product Manager
sarah.kim@email · linkedin.com/in/sarahkim
EXPERIENCE
Stripe — Product Manager 2022–Present
  • Leveraged cross-functional synergies to drive impactful stakeholder outcomes
  • Spearheaded initiatives that enhanced organizational capabilities across multiple verticals
  • Demonstrated strong leadership skills and a growth mindset
  • Passionate team player with excellent communication abilities
SKILLS
Leadership · Communication · Problem-solving · Teamwork · Strategic thinking
→
Merl · Editor ✓
Sarah Kim
Senior Product Manager · Fintech · B2B Payments
sarah.kim@email · linkedin.com/in/sarahkim
EXPERIENCE
Stripe — Product Manager, Merchant Dashboard 2022–Present
  • Launched 3 merchant-dashboard features; grew weekly active merchants 22% in 6 months
  • Cut onboarding drop-off 31% by rebuilding the KYC flow with design + risk teams
  • Shipped B2B invoicing v2 end-to-end with 4 engineers; $18M ARR in first quarter
  • Led fraud-modeling workstream; reduced chargeback rate from 0.9% → 0.4%
SKILLS
SQL · Amplitude · Figma · Ruby on Rails · Fraud modeling · Experimentation
Why Editor wins: Same work history, zero invented metrics. Editor's skill file runs keyword SEO + GEO against the posting — extracts the exact vocabulary a human recruiter AND the ATS parser look for (B2B payments, fraud, onboarding, KYC), then mirrors those terms using your real experience. Skills row rewritten from "communication" to actual tools the JD asked for. Reads clean for the hiring manager, scores high for the bot.
Task · LinkedIn post about a work win
Generic AI ✗
SK
Sarah Kim · 1st Senior PM at Stripe · ex-Figma · hiring 2h · 🌐

🚀 Thrilled to share a milestone! This quarter, I had the privilege of leading a cross-functional initiative that leveraged synergies across multiple verticals to drive impactful stakeholder outcomes. Key takeaways:
(1) Collaboration is key
(2) Embrace a growth mindset
(3) Data-driven decisions win

Grateful for the journey and excited for what's next! 🙌 #leadership #growth #innovation #pm

👍 Like💬 Comment↻ Repost➤ Send
→
Merl · Echo ✓
SK
Sarah Kim · 1st Senior PM at Stripe · ex-Figma · hiring 2h · 🌐

Cut our pitch deck from 40 slides to 12 this weekend.

The first 28 were context nobody asked for. Stakeholders didn't need the backstory — they needed the ask, the number, and the timeline.

Result: 22% higher close rate this quarter on the same pipeline.

Reminder to myself — clarity is the product. Say the thing.

👍 Like💬 Comment↻ Repost➤ Send
Why Echo wins: Echo trained on 3 of Sarah's actual posts. Lowercase. No emojis. No "thrilled to share." Specific number. Her actual self-deprecating tone. That's why her connections stop scrolling — generic AI posts blur past in two seconds.
Task · LinkedIn headshot from a handful of front-camera selfies
Raw selfie · before
Raw selfie before Shine
→
Merl · Shine ✓
Shine headshot after
Why Shine wins: Same face. Same bone structure, same eyes, same smile — but lit properly, framed for LinkedIn, wearing something that says "hire me" instead of "just woke up." Shine's face-identity filter rejects ~90% of raw generations before you ever see them. You only get the ones that still look like you.
Task · UGC-style product ad — a creator pitching a lotion
Generic AI ✗
→
Merl · Cutter ✓
Why Cutter wins: Listen to the raw clip ↑ — hands do weird stuff, product warps, voice sounds like a robot's approximation of a person. Cutter routes every shot to the best video model in the world for that specific shot, runs a quality-filter pass for face/hand warps, and re-rolls anything that doesn't make the cut. You post what you'd actually post.
Task · restyle a cramped dorm room — keep the space, change the soul
Your room · before
Room before Designer
→
Merl · Designer ✓
Room after Designer
Why Designer wins: Same windows, same light direction, same floor plan — intact. Designer extracts the architecture first, then restyles inside the bones you've got. New palette, new furniture, new soul — but still your room. You could actually buy this.
· Made with Merl ·

What people are making.

Every clip and shot below came out of Merl on the first try — because we trained each agent with thousands of lines of skill files, taste rules, and reject filters. The models give the horsepower. We gave them the eye.

· Why Merl ·

General AI is good at everything. Not exceptional at anything.

The problem with one big AI.

Ask any general AI for a résumé and you get... a résumé. Technically correct. Zero personality. Hiring managers can smell the model three bullets in. That is the ceiling of "one AI for everything."

So we took the best models.

We took the best AI models in the world — Claude for reasoning, Google's Nano Banana 2 for image, frontier video — then we trained them. Thousands of lines of skill files, taste rules, examples, counter-examples, reject rules — per agent. The models supply the intelligence. We supply the craft.

And got really, really good at 7 things.

Résumés that land. Headshots that keep your face. Video that doesn't warp. Tweets and captions in your voice. Essays rewritten without the AI smell. Study breakdowns that actually stick. Rooms restyled with the bones intact. We only ship when they hit the bar.

· Pricing ·

Pay what you use. Or don't — try first.

One free cast per agent. Top-up mana never expires.

Day Pass $4.99/ 24h

An afternoon of work — a résumé pass, a cover letter, some headshots.

  • 75 mana
  • Every agent unlocked
  • No renewal
Get day pass
MOST POPULAR Pro $19.99/ month

Video + everything, for people who actually use it.

  • 400 mana / month
  • Best-in-class models, trained by us
  • Priority support
  • Every agent
Get Pro →
Weekly $9.99/ week

Job-hunt mode. Enough for a full application round — résumé, letters, headshots.

  • 175 mana / week
  • Every agent
  • Cancel anytime
Get weekly
· FAQ ·

Questions, answered.

Aren't you just using the same AI as everyone else?
We run the best AI models in the world — Claude for reasoning, Google's Nano Banana 2 for image, and frontier video models for clips. Frontier labs spend billions training those — we're not gonna compete with that, lol. What we will do is take those models and train them on a few specific jobs. Hand-written skill files. Real output filters. Years of taste baked into every prompt. The models are the engine. We tuned it. That's what makes a résumé land, a headshot look like you, and a video not warp faces.
Do you train on my data?
No. Ever. Your inputs aren't used to train any model, ours or anyone else's. Your scrolls (history) live on your device unless you choose to share.
What does mana get me?
Depends on the agent. Text agents (Editor, Echo, Essay, Study) are typically ~5 mana per cast. Image agents (Shine, Designer) are ~12. Video (Cutter) is 30–60 depending on length. Mana never expires.
Can I cancel anytime?
Yes — managed through your Apple ID subscriptions. Cancel whenever, keep your remaining mana for that billing period. No "hit us up on email" traps.
Why specialists instead of one big AI?
Because "one big AI that does everything" falls back on the average. A résumé specialist has spent its whole life thinking about résumés. You wouldn't hire a generalist to cut your hair — same logic.
Does the free version actually work?
Yes. One free cast per agent — no trial timer, no credit card required. Try the whole crew, decide if any of them earn a spot in your pocket.

The last AI app
you'll actually use.

Free first cast per agent. iOS, Android, or straight in your browser — take your pick.

iOS → Android → Web →