01 / 03
Customer-facing agents
Website chat and in-product assistants wired to your real orders and accounts, with hard limits on what they can do without a human.
Like Ranger, below.
// AI ENGINEERING FOR FOUNDERS & PRODUCT TEAMS
Get your AI feature
into production.
And keep it there.
We build and operate AI features inside your existing systems. Bring us a scoped project or an ongoing roadmap.
14 years shipping production software · $300M+ traded through Pear, a platform we helped build · 600+ bookings through KaribuKit, our own hotel software
// WHAT WE BUILD
The demo took a weekend. Production is where it stalls.
Real customer data, permissions, edge cases, cost, and the failure modes a prompt can't see. That's the work we take on.
02 / 03
Internal workflow agents
Ops, quoting, scheduling, back-office. The work your team does by hand today, with a person approving the steps that matter.
Like the staff copilot behind Mara Hilltop, below.
03 / 03
Decision systems with guardrails
Pricing, trading, routing. Language models where they help, deterministic code wherever money or correctness is on the line.
Like Agent Pear, below.
// PRODUCTION WORK
One built for a client. Two we run ourselves.
All three are live today. Click through to any of them.
Mara Hilltop · our safari lodge · runs on KaribuKit
An AI agent that answers safari guests around the clock.
Ranger handles inquiries on the lodge website, quotes from live availability in the guest's language, and hands staff a ready-to-confirm booking. Staff run reservations, check-ins and pricing by chatting with the same agent in the app or on Telegram. Confirmation and payment stay with a human by design.
We built KaribuKit, the property system underneath. 600+ bookings have run through it since March, and we live with every edge case.
Visit ↗ KaribuKit PMS · custom MCP server · multi-model
R
Ranger
Mara Hilltop · sample conversation
DEMO
Pear Protocol · Agent Pear · client
A trading agent that sets up the position, then waits for the trader to confirm.
Agent Pear answers pair and market questions from live on-chain stats and sentiment, then stages long/short positions that execute only after the trader confirms. Live for Pear's retail traders on Telegram and web.
Two years as Pear's engineering partner. $300M+ has traded through the platform we helped build, and the on-chain vaults we built end to end filled their first $1M cap about ten hours after launch.
Visit ↗ Multi-model DAG · on-chain + sentiment pipeline · confirm-before-execute
Agent Pear · Signal
Sentiment divergence +2.4σ · trade staged, awaiting confirm
SlopIt · publishing · ours
A blog platform where the AI agent is the user.
An agent publishes a post with one API call and gets a live URL back. Every post on our blog ships through it.
slopit.publish({
title: "Why remote work wins",
body: "...",
tags: ["remote", "work"]
}) → 200 OK · 142ms
Live at nj.slopit.io/why-remote-work-wins
// HOW IT WORKS
Forward-deployed engineers.
In your repo, your systems, your standup.
01
A real conversation, free
45 minutes with NJ on what you're building and where it's stuck. If AI isn't the right fix yet, he'll say so.
02
Scope it, then ship it
We agree on one workflow and what done means, then build it in your repo for a fixed price.
03
We keep going, or we don't
If it works, the next piece on your roadmap. If it doesn't, you're not locked into anything.
// PRICING
Clear pricing. No discovery call to find out what it costs.
For scale: one senior AI engineer in-house costs $300K+ a year loaded, once you find one.
Start here
AI Build Sprint
$12,500
Two weeks to a working AI feature in your environment. One workflow, agreed acceptance criteria, in your repo. 30 days of support after launch.
Bigger builds, like several systems or complex permissions, get their own quote on the first call.
If we can't see a clear first build on that call, we'll say so and won't take the sprint.
Start a sprint →Embedded AI Team
$10,000/mo
Two engineers embedded with your team, NJ on technical direction. We own what we've shipped, watch it in production, and work your backlog every month. Two-month minimum, cancel anytime after.
Already have a roadmap? Skip the sprint and start here.
Talk to us →// THE TEAM
No account managers. No layers. You work with the people who build.
Founder & Chief Architect
NJ
NJ has spent 14 years building software for other people's businesses, including Pear's trading platform, and now AI agents. He also runs a safari lodge in Kenya on software our team built. And he picks up the phone.
Direct line California · +1 (717) 683-9393 · call or WhatsApp
Senior Engineer & Delivery Lead
Deep
Leads client delivery end-to-end. The person who makes sure your system works in production, not just in a demo.
Senior Engineer & Systems Architect
Pranjal
ex-Swiggy. Our go-to for the hardest problems: Web3 infra, AI systems, anything where “figure it out” is the spec.
Full-Stack Engineer (Design Lead)
Chetan
Full-stack with design instincts. Most of our UI passes through him before it ships, and more and more of the AI integrations underneath it.
Full-Stack Engineer (Backend / Performance)
Hemanshu
Backend-first full-stack. Built a 250 req/sec layer behind one of our hospitality systems. Owns caching, API design, the parts that have to scale under load.
// VIDEO & WRITING
I show the work.
Short videos of AI doing real jobs, start to finish. The long version goes on the blog.
I think I've solved procrastination
Indexing a year of video locally on a 2021 MacBook. Front page of Hacker News, 470 points.
// LET'S BUILD
Tell us what you're building.
A few lines on what you need in production. NJ reads every one and replies within a business day, including when we're the wrong team for it. Need a written proposal for your team first? Say so.
Or skip the form Call or WhatsApp NJ at +1 (717) 683-9393. He picks up.
// Thanks, received
We'll be in touch within one business day from nj@simbastack.com.
// Couldn't send
Something went wrong on our end. Email
nj@simbastack.com directly and we'll pick it up.