01Founders
You want to launch your startup, but you don't have a tech team yet.
We validate the idea and put the first version live.
Who Kapstan works with
02Companies and specialists
You know your field from the inside, from law to construction, and want to create a digital product. Software and marketing aren't your business.
You bring the business and we build the rest, from prototype to maintenance.
03Tech teams
Your team already delivers, but no one is free to drive a new AI track.
An AI PM works inside your team, 1 or 2 days a week.
Does this sound familiar?
“I have an idea, but I don't know if it's worth the investment.”
We talk to real customers, build a clickable prototype and test it before spending on development.
01 · What you can build
- A clickable prototype of the app, to show an investor or your first customer
- A waiting page with a sign-up list, to measure demand before building
- A price test: the offer in the hands of real customers, to see what they pay
- The scope of the first version: what goes in now and what waits
Who it's for Founders · Companies and specialists before raising, or before opening a new revenue line.
Discover · from US$ 600 This is my case
“I need to launch the first version fast, without building a team.”
Your MVP live in weeks, with the code being yours.
02 · What you can build
- A marketplace that connects supply and demand in a niche
- A subscription product, with sign-up, recurring billing and a dashboard
- An on-demand service app: order, payment and tracking
- A course or content platform with a members area
Who it's for Founders to reach your first paying users without hiring a team.
Build · from US$ 3,000 This is my case
“My customers still buy and book through WhatsApp and spreadsheets.”
Your own sales, scheduling or booking system, with your brand.
03 · What you can build
- Your own store with catalog, cart, Pix and card
- A customer portal: orders, invoice copies, history and documents
- Booking or scheduling with payment up front
- Online quotes: the customer builds the order and gets the price right away
Who it's for Companies and specialists distributors, manufacturers, schools, consultancies, tourism and services.
Build · from US$ 3,000 This is my case
“Leads come in, but they go cold before anyone replies.”
A funnel that answers right away with your company's information, qualifies, and passes on to sales only who is ready.
04 · What you can build
- A WhatsApp agent that answers price, lead time and stock with your company's data
- A lead qualifier: it asks, scores and hands the rep a summary
- An internal assistant that answers the team's questions about processes and contracts
- Quote and follow-up: the agent drafts the next message, the rep approves and sends
Who it's for Companies and specialists those who already generate demand and lose sales in the wait between contact and reply.
Build · from US$ 3,000 This is my case
The agent that answers and books at 11pm on a Sunday
To sales, only the lead that is ready
“I don't know where my best customers come from, or why the others leave.”
A dashboard that shows which channel brings who buys and where you lose who gives up.
05 · What you can build
- The origin of each sale: channel, campaign and cost per customer
- The funnel step by step: where people stop between the visit and the purchase
- Retention by cohort: who comes back, who disappears and when
- The daily operation that lives in a spreadsheet today: orders, stock and delays on one screen
Who it's for Companies and specialists those who invest in marketing and operations and decide without seeing what came back.
Build · from US$ 3,000 This is my case
“We want to put AI in our product, but the team is busy with day-to-day delivery.”
An AI PM who finds where AI creates value and delivers alongside your team.
06 · What you can build
- Search and recommendation that understand what the user meant
- An assistant inside the product that carries out tasks for the user
- Summary and extraction from documents, calls and tickets
- Onboarding that adapts to what each customer uses
Who it's for Tech teams a full roadmap, pressure to ship AI, and no one free to find where it pays off.
Lead · from US$ 1,600/mo This is my case
“My product is live, but it doesn't grow.”
Acquisition, activation and retention built inside the product, with short, measured experiments.
07 · What you can build
- A landing page with tracking, to know what each visit turned into
- Referral with a reward: the customer brings the next customer
- Activation: the first steps that take a new user to the first result
- Win-back: e-mail and personal messages with the right reason to come back
- A/B tests on price, sign-up and offer
- Growth metrics inside the product itself, from day one
Who it's for Founders · Tech teams product live, users arriving, and the curve flat. This is what the market calls GTM engineering.
Build or Lead · we decide in the call This is my case
Not just software, not just marketing
Software houses build well, but they rarely think about how the product will grow. Marketing agencies think about growth, but they don't build product. Kapstan does both.
Writing code got cheap, and the team only large companies could afford now fits in a small crew. What stays expensive is deciding what to build, and for whom.
| Software house | Marketing agency | kapstan | |
|---|---|---|---|
| Finds out what's worth building | Rarely | No | Yes |
| Builds product that works | Yes | Rarely | Yes |
| Product is born measuring results | Rarely | Sometimes | Yes |
| Thinks about customer acquisition and retention | No | Yes | Yes |
| AI applied in the product | Sometimes | Sometimes | Yes |
| Works inside your team | No | No | Yes |
The Kapstan Kapstan
The product, engineering and growth team a Series B startup has, working as an extension of your company.
See how your business gets among the best
Today anyone can ask an artificial intelligence agent for an app, and get one.
Tools like Lovable, Bolt and Cursor write an entire app from a single sentence. Except your competitor made the same request and got the same product: the agent answers with the average of what it has already seen.
Nobody gets among the best with the average.
What takes your business out of it is what the agent doesn't have: your customer, good taste and a product thought through together with marketing from day one.
-
Design Thinking · positioning
We understand your customer before drawing a screen
We interview the people who buy from you and those who gave up, map the market and the competitors, and define who the product is for, what problem it solves better than the alternatives and how much that customer is worth. It's the foundation for everything that follows, from the product to the ad.
What you get
- Interviews with customers and with those who gave up, with what they said organized by theme
- The ideal customer profile (ICP): who buys, who decides and who is worth serving first
- The job the customer hires the product to do, in their own words
- The map of competitors and alternatives, and the position where none of them is strong
- The positioning and value proposition, in the sentence that goes on the site and in the ad
Part of Discover
-
Lean Startup · validation
We test the idea with customers before investing in development
A clickable prototype and a landing page go in front of real people. We measure whether the promise attracts, how much each lead costs and how much they would pay, and only then decide what goes into the first version.
What you get
- A clickable prototype, tested with five to eight people from your audience
- A landing page with a test ad: the cost per click (CPC) and the cost per lead, measured before the product exists
- A price test with a real offer, to learn what the customer pays
- The waitlist with the first interested people, ready for launch
- The scope of the first version (MVP): what goes in now, what waits, and why
Part of Discover
-
Agile methods · engineering
We build software with the rigor of a senior engineering team
Test-driven development, continuous delivery and human review of all code written by AI agents. The design has the finish of a benchmark product, and the code is yours from day one.
What you get
- Tests written alongside the code (TDD), running on every change
- Continuous delivery: every improvement goes live as soon as it's approved, without stopping the product
- Human review of all code written by AI agents, before it gets in
- Design and copy reviewed until the product has the finish of a benchmark in your market
- Repository, infrastructure and accounts in your company's name, with documentation
Part of Build
-
Growth · acquisition and activation
We put the acquisition funnel inside the product
Acquisition and activation are part of the product, not a campaign after it. Every channel is tracked from the first visit to payment, and the first use is designed to get the customer to a result as soon as possible.
What you get
- End-to-end tracking: which channel and which campaign each customer came from
- The customer acquisition cost (CAC) per channel: search, content, ads and referrals
- SEO and presence in AI assistants' answers, to be found without paying per click
- The onboarding that takes a new user to their first result, with the activation moment measured
- A referral program inside the product, so users bring the next customer
Part of Build
-
Data analysis · experiments
We measure the whole funnel and improve what weighs most on revenue
The whole funnel sits on one dashboard: acquisition, activation, retention, revenue and referral. Each cycle we tackle the stage that weighs most, with an experiment whose result is measured, until the business is among the best in its market.
What you get
- The pirate funnel dashboard (AARRR): acquisition, activation, retention, revenue and referral
- CAC against LTV: what it costs to bring in each customer, and what they spend over time
- Cohort retention: how many customers from each month are still active
- A/B tests with measured results, one per cycle, on the stage that weighs most
- Monitoring and maintenance, with the alert arriving before the customer complains
Monthly maintenance, after Build
For technology teams, the same fronts happen inside your team, in Lead.
The Kapstan lab
It's where Kapstan tests today, on its own product, the method it uses on yours. It's all open for you to see.
Cidade.ai + AgileHarness
An app built by AI agents, and you can follow along
The Cidade.ai runs entirely on WhatsApp: it suggests outings, proposes challenges and connects you to professionals. It is the second time I solve this problem; the first was Alegreme, in 2019, with machine learning. It is built by AgileHarness, our system in which AI agents plan, code and test, with review at the critical points.
Oficina · plugins and skills
AI agents ready for work
The Oficina brings together plugins and skills for Claude Code, with agents by profession. The prospecting one turns your WhatsApp into a CRM; the job-search one organizes your job hunt and builds a résumé for each opening, without making anything up.
- ProspectingWhatsApp becomes a CRM
- Job searcha résumé per opening, nothing made up
- The next professionunder construction
Who is behind Kapstan
Jonatas Salgado
founder and AI Product Manager
For 14 years I've worked between marketing, data, code and product, and since 2019 with AI. I was always the product manager who also coded and also looked after acquisition.
With AI agents writing code, that stopped being a pile of job titles and became a way to deliver: whoever decides what to build also builds it, and also measures it.
I created Kapstan to do that alongside whoever has the idea and the market.
Kapstan isn't an agency: it's me, with partners assembled per project. Do you work with design, video or code? Join the network.
Experience at
14 years in product
-
Warreninvestment fintech · Head of Product Marketing, then Product Manager (Partner)
I built marketing from scratch along with the whole acquisition and activation funnel: affiliates paid per activated customer, in-app friend referrals and automation. In that period the company went from seed to Series A, with Kaszek and Ribbit, both Nubank investors.
150 → 50kcustomers
-
Antlerventure capital · Founder in Residence · Cubo Itaú, São Paulo
I was selected for the founder residency at Antler, a global venture capital fund, where I formed teams and validated businesses on four fronts: e-commerce, community marketing, market data forecasting and autonomous AI agents for the investment sector.
Under 1%of applicants are accepted
-
ThoughtWorkssoftware consultancy · training program
I learned to code with ThoughtWorks consultants, at Tecnopuc: agile methods, TDD, continuous integration and pair programming, in Python, TypeScript, Java and Ruby. All by hand, before AI, and that is why today I know how to review what an agent writes.
Developersoftware
-
Microsoft Innovation CenterPUCRS innovation center · Tecnopuc
I was scrum master and agile mentor to cross-functional teams, from concept to delivery on every project, with Lean and Design Thinking in all of them, alongside ThoughtWorks professionals.
Scrum masterand agile mentor
-
ACE Cortexinnovation consultancy
I led the rollout of agile methods and design thinking in a digital transformation project in fashion retail, with the business and technology areas deciding together, in short cycles.
Lojas Rennerthe project's client
-
WPPHogarth · pharmaceutical · freelancer
For AstraZeneca and Galderma, Hogarth's clients, I measured how doctors used the two companies' platforms. It started as reports built from Google Tag Manager and became an automated system that reached the sales team ready to use.
Data scientistPython, SQL and Data Studio
-
Ventiurstartup accelerator
I was an AI and software mentor in more than one acceleration program, with startups at two moments: early stage, deciding what to build first, and traction, when the product has to grow without breaking.
20 startupsmentored
-
Órulonew-build homes for real estate agents · mid-level developer
I helped build órulo.com.br, which indexes new-build homes for real estate agents, and supercambio.com, a peer-to-peer currency exchange platform.
Ruby on Railsand Angular
-
Cadastradigital marketing · marketing analyst
I ran the Livraria Saraiva digital account: SEO, analytics and return on investment, focused on sales performance.
R$ 20ka day in media
AlsoArvo Consultoria, from 2020 to 2026, which became Kapstan in 2026
Three ways to start
-
Discover
To know what's worth building before you invest.
A quick diagnosis or a clickable prototype tested with real customers.
From US$ 600
Learn more -
Build
To put your product live and keep it running.
App, system or MVP with metrics from day one.
From US$ 3,000
Learn more -
Lead
For teams that want to accelerate an AI track.
A fractional AI PM, 1 to 2 days a week, in the forward deployed model.
From US$ 1,600/mo
Learn more
Equity partnership
Make Kapstan your technology partner
For those who know the market, have the customer and want to build the product with someone who knows product and technology.
Common questions
How much does it cost and how long does it take?
Discover starts at US$ 600 and takes about a week. Build starts at US$ 3,000, with a timeline that follows the scope. Lead starts at US$ 1,600 per month. The exact price comes after the call, with scope and deadline agreed in writing.
Does the code stay with me?
Yes. The repository and the accounts are in your company's name from day one.
What if a problem shows up after delivery?
We fix what is part of what was agreed. After that, you choose between monthly maintenance with Kapstan or carrying on with your own team.
Will I depend on Kapstan?
No. Code, documentation and decisions are recorded in a way that any team can pick up.
Do I need to understand technology to hire you?
No. You bring the problem and the knowledge of your market. Technical decisions are explained in plain language, without jargon.
Do you work outside Porto Alegre?
Yes, across Brazil and abroad, working remotely. In Porto Alegre, in person as well.
How does the equity partnership work?
There is 1 slot per quarter. You apply through the form and we assess case by case. There is always a cash portion, and the rules are set in a contract.
Does it integrate with the systems I already use?
In most cases yes — WhatsApp, payment providers, management systems and spreadsheets. We confirm it during the diagnosis, before any promise.
Do you do GTM engineering, or work as forward deployed?
Both, without using the name. Lead is the forward deployed model: whoever builds works inside your company, next to whoever uses it. Growth built inside the product is what the market calls GTM engineering. The guides explain what each term names: GTM engineering and forward deployed engineer.
