What does the AI Readiness Audit deliver?

One to two weeks. Workflow audit, prioritized opportunity roadmap, ROI projection, and a working demo of one AI capability on your real data. Not a slide deck.

Do you work with apps built using Lovable, Cursor, or Bolt?

Yes. Vibecoder graduation is one of the most common engagements. We take a prototype built in Lovable, Bolt, Cursor, or Replit and harden it for production: auth, secrets, eval suites, monitoring, model routing. Code lands in your repo from day one.

Will I get junior people on my project?

No. Strategy, design, and engineering run through senior leadership on every engagement. The person who scopes the work writes the first commit. No account managers, no junior staff handing off to other junior staff.

How is Kastling different from a generic AI agency?

We build AI products of our own (Verdikt is live in alpha, with Capsule, Lanes, and Triage in development). We bundle brand and engineering under one team, not separate agencies. Code lands in your repo, evals are the contract, no token arbitrage, IP transfers at engagement end. Most agencies do none of these.

How does pricing work?

Every engagement is scoped to outcomes and quoted after a short call. No hourly billing and no public price list. Book a call and we will scope it with you.

What's an alternative to hiring a full AI team in-house?

A Project or Partnership with us. A senior AI hire is six to nine months to onboard and an annual cost in the high hundreds of thousands. A Project ships a complete AI capability in 4 to 12 weeks on a fixed, scoped engagement. A Partnership covers the same outcome over time as a monthly retainer with a documented exit ramp.

What if scope changes mid-engagement?

Small adjustments we absorb. Material scope changes get a written change order with a new fixed price. No mystery invoices in week eleven.

No. We're a studio, not a venture studio. Cash engagements only. Keeps incentives clean and the work shipped on time.

Where are you based, and do you work internationally?

Distributed studio. We run engagements across all major time zones. Pricing is in USD. We do not discount by geography.

What tools and models do you work with?

Model-agnostic by architecture. A thin routing layer (LiteLLM or equivalent) sits between application code and any model provider. We use OpenAI, Anthropic, Google, and open-weight models as the task demands. Tooling stack includes Figma, Next.js, React Router, Tailwind, Vercel, plus full evaluation stacks like Braintrust, Promptfoo, or custom eval harnesses.

How fast can you start?

Most engagements kick off within 7 to 14 days of the scope call. Audits start sooner. We do not pad the timeline and we will tell you on the scope call if we can't start when you need us to.

You. Full IP transfer on final invoice. Every prompt, every eval, every fine-tune, every piece of routing logic is yours. No license games, no usage caps, no portfolio rights without your sign-off.

Standard mutual NDAs, yes. Send it with the brief and we sign before the scope call. Bespoke or one-sided NDAs we'll negotiate.

What's 'evals as the contract'?

Before we write code, we define what 'good' looks like as a set of inputs, expected behaviors, and quality thresholds. That eval suite lives in your repo. CI fails the build when thresholds slip. When a model upgrades or a prompt changes, the evals tell us whether quality moved. We do not ship a feature whose eval suite is empty.

What is 'no token arbitrage'?

You pay model providers (OpenAI, Anthropic, Google) directly through your own accounts. Costs land in your billing, not ours. We surface per-feature cost dashboards weekly. The unit economics are yours to audit and optimize.

Can I see examples of past work?

Verdikt is live in alpha at tryverdikt.app, and the Shipped section of the homepage walks through how we built it. Client case studies go up as engagements wrap and we have permission to publish. Ask on the scope call and we'll share relevant references that match what you're building.

What are Kastling's in-house AI products?

Verdikt (live, alpha): a verdict on a startup idea before you build it. Capsule (in development): a daily outfit from your own wardrobe. Lanes (in development): one keystroke to switch your Mac between projects. Triage (in development): turn a brain dump into a calm day plan. We build these to keep the studio's methodology honest.

What platforms do Kastling's apps run on?

Verdikt is a web app. Lanes is macOS native. Triage is Mac and iPhone, on-device. Capsule is iOS. The Shipped section of the homepage has the specifics.

Where does my data go when you build something for me?

Into your accounts, on your infrastructure, in your repo. We do not retain client data on our systems beyond what the engagement requires. Model providers see only what your application sends them, billed to your accounts. For sensitive workloads, we configure zero-retention model endpoints by default.

We design, build, and launchAI products.

Kastling is the AI Studio. New products from zero, or AI inside the systems you already run. Verdikt is live in alpha, with three more in development. Now we ship yours.

See what we offer

It starts with an AI Readiness Audit. One week to a roadmap and a working demo on your data.

Service pillars

Day 1

Code in your repo

100%

IP yours at the end

Where you start

Two ways in. One studio.

Start a new product from zero, or put AI to work inside the systems you already run. Same team, same playbook.

From zero

Launch a new AI product.

You are a founder taking an AI product from a blank page to a live, paying product.

Brand, build, ship, and operate under one team.
Investor-ready narrative and a working demo on your data.
Vibecoded prototype hardened to production.
Eval suites and monitoring wired in from day one.

Inside your stack

Add AI to your operations.

You run a business and AI fits somewhere in your operations. You want ROI, not a slide deck.

An AI Readiness Audit that ends in a roadmap, not a maybe.
Chatbots, workflow automation, and RAG assistants in production.
Built on your stack, your data, your guardrails.
Handover with the eval suite that proves it works.

What we offer

Everything an AI product needs. Under one roof.

Strategy, engineering, brand, and operations. Take all four, or just the part you are missing.

AI Strategy

Diagnose before you build. Where AI fits, what it is worth, and what good looks like.

AI Readiness Audit: workflow, tech, and opportunity mapping
AI product strategy: positioning, ICP, value architecture
A 90-day roadmap with prioritized use cases
Investor narrative and business case
Evaluation framework design: what good looks like, defined up front

Apps

We build AI products of our own. Now we build yours.

Verdikt is live in alpha. Three more in development.

01
Verdikt
Live, alpha.
A verdict on your idea before you build it.
Describe your idea. Verdikt's agent interviews you on the details, runs deep research across competitors, market, and audience, then delivers an analyst-grade report in hours, not weeks.
02
Capsule
In development.
A daily outfit from your own wardrobe.
Open the app, see today's outfit pulled from clothes you already own. The taste of a stylist, the patience of a friend, the read on your closet nobody else has. Ask it before you get dressed. Ask it before you buy.
03
Lanes
In development.
One keystroke to switch your whole Mac between projects.
Apps, docs, browser tabs, window positions, all of it. Project A to Project B in one shortcut. Your context, restored.
04
Triage
In development.
Dump everything in your head, get a calm day plan in return.
Mac and iPhone, on-device. Capture the chaos in one minute, leave with a plan you can actually run today.

One email per launch. No newsletter, no nudges.

How we engineer

Seven principles. No exceptions.

What we promise, how we deliver, and what goes in every contract we sign.

01
We build our own
Verdikt is live in alpha. Capsule, Lanes, and Triage are next. We bet on our own work before we sell yours.
02
Code in your repo
Production code lives where you will maintain it. No vendor sandboxes, no deliverables thrown over the wall.
03
Evals are the contract
We define what good looks like before we write code, then ship the eval suite alongside the feature.
04
Model-agnostic by architecture
A routing layer sits between your code and any provider. No lock-in to OpenAI, Anthropic, or Google.
05
No token arbitrage
You pay model providers directly. We never mark up tokens.
06
Brand and engineering, one roof
Ship the product and the launch from one team. No re-briefing a second agency after the build.
07
IP transfer at the end
Every prompt, every eval, every fine-tune is yours when the engagement closes.

See what we offerAll seven, in every contract we sign.

Process

Four steps. Brief to ship. No pitch theater.

3 min
01
Brief
You submit a project brief. Three minutes. We read every one and reply within one business day.
30 min
02
Scope call
30 minutes. We define the engagement shape (Audit / Project / Partnership / Embedded) and what done looks like. No discovery deck. No retainer pitch.
48 hrs
03
Proposal
Written proposal in your inbox within 48 hours. Scope, timeline, fixed price, eval criteria, IP terms. Sign with one click.
Week 1
04
Ship
Work begins within the week. Code lands in your repo from day one. Final deliverable on the date we agreed. Optional Operations retainer. Never a lock-in.

Fit

We work best with operators who already moved.

Right call if

You're a founder launching an AI product. You want brand, build, and ship under one team.
You shipped a prototype with Lovable, Cursor, or Bolt. The production version is where it falls apart.
You're an SMB or mid-market business and you know AI fits somewhere in your operations. You don't want a twelve-week PDF.
You raised and you want the seed round to ship a product, not a consultancy roadmap.

Wrong call if

You want a strategy deck with no shipped code. We don't do that.
You want pure offshore staff augmentation by the hour. We don't do that either.
You want the cheapest option on the internet. Fiverr is on the internet.

FAQ

The questions every founder asks before sending a brief.

An AI studio. We design, build, and launch AI products end-to-end across four pillars: Strategy, Engineering, Brand & Launch, and Operations. We serve founders launching new AI products and businesses adopting AI inside their operations.

Start an audit

Tell us what you are building. We will tell you if we can help.

A brief takes three minutes. We read every one. If there is a fit, you hear back within one business day with a scope call and a proposal. If there is not, we say so and point you somewhere better.

Email the team

Code in your repoEvals as the contractModel-agnosticNo token arbitrageIP yours at the end

We design, build, and launchAI products.

Two ways in. One studio.

Launch a new AI product.

Add AI to your operations.

Everything an AI product needs. Under one roof.

AI Strategy

We build AI products of our own. Now we build yours.

Verdikt

Capsule

Lanes

Triage

Seven principles. No exceptions.

We build our own

Code in your repo

Evals are the contract

Model-agnostic by architecture

No token arbitrage

Brand and engineering, one roof

IP transfer at the end

Four steps. Brief to ship. No pitch theater.

Brief

Scope call

Proposal

Ship

We work best with operators who already moved.

The questions every founder asks before sending a brief.

Tell us what you are building. We will tell you if we can help.