# AltSlate Labs

> AltSlate Labs builds and runs agent-first software for growing companies. AI agents carry out the routine operations of the business; its people set the rules, supervise, and decide what matters.

Last updated: 2026-10-06. Contact: hello@altslate.com. Website: https://www.altslate.com

AltSlate Labs is an AI engineering and software services firm led by its founders, who work with each client as one senior pod. It serves mid-sized companies that need more capacity without adding headcount at the same pace.

**What "agent-first software" means.** Most software adds an AI assistant to existing screens. Agent-first software is built the other way round: AI agents are the operating core. A supervisor routes each request to a specialist agent (scheduling, bookings, billing, customer conversations, outreach, reporting, payments), the agent does the work, and the team directs the agents through dashboards and mobile apps.

**How people stay in control.** Oversight is set by the kind of work:
- Finance, accounting and regulated work: a person approves every change before it is applied.
- Day-to-day operations: agents act within limits the client sets; exceptions go to a person.
- Marketing, outreach and sales: agents act; people supervise and review results.
- Each agent reaches only the records its role allows, and every action is recorded.

**Outcomes delivered to clients.**
- 37 AI agents working together handle more than 80% of one client's daily tasks.
- Accounting close cut from 10 business days to 2.
- Reconciliation of records cut from days to hours.
- Tasks generated from real-time field data in under 60 seconds.
- Transaction risks gathered and flagged across three continents in under 1 hour.
- Four AI products taken from concept to paying customers on three continents in under 12 months, and operated since.

**How to start.** Start with one problem. Timelines are fixed and each engagement ends with something the client owns.
- Quick Build, 4–8 weeks: one problem scoped, built, tested with the client's team and handed over running.
- Architecture Review, 2–4 weeks: an outside opinion on existing software or a design.
- Product & Customer Discovery, 4–6 weeks: a customer pain turned into a prioritized product plan.
- Full-Platform Partnership, 3+ months: a full platform with agents at its core, built and then run with the client.

**Also does.** Real-time data platforms, legacy software modernization, voice and conversational AI, and training compact language models (the open Jugnu research family).

**Industries with shipped work.** Energy and upstream oil and gas, financial services, healthcare, consumer marketplaces, climate and agriculture, and AI-native software.

## Main pages

- [Home](https://www.altslate.com/): Agent-first software, client outcomes, how oversight works, and how to start
- [Portfolio](https://www.altslate.com/portfolio): Case studies, browsable by industry
- [Models](https://www.altslate.com/models): The Jugnu family of compact language models and published research
- [Blog](https://www.altslate.com/blog): Engineering and product articles
- [Team](https://www.altslate.com/team): The founders and their background

## Case studies

- [An Agent-First SaaS Platform, With Humans in Command](https://www.altslate.com/portfolio/agent-first-saas): A venture-backed client wanted to scale without hiring at the same pace. About 37 AI agents now carry the routine operational work, and the client chose to have a person approve every change before it applies. Live in production, with web and mobile apps for staff.
- [Turning Ambitious AI Visions into Production Reality](https://www.altslate.com/portfolio/us-001): A consumer trust-and-safety company asked us to take its AI product line from concept to market. We designed, built and now operate four AI products, live with paying users on three continents in under a year.
- [AI Chatbot for Remote Patient Monitoring](https://www.altslate.com/portfolio/rpm-chatbot): Growing a remote patient monitoring program used to mean growing the nursing team just as fast. Our AI layer handles routine check-ins and surfaces urgent cases, so a fixed care team can watch a larger panel and act sooner.
- [Real-time Production Accounting for Oil & Gas](https://www.altslate.com/portfolio/production-accounting): An upstream oil and gas operator took ten business days to close each month. We rebuilt production accounting: the close takes two, the monthly allocation run fell from eight days to a few hours, and field teams see daily results.
- [Equine Marketplace & Social Platform](https://www.altslate.com/portfolio/equine-marketplace): Equestrian buyers, sellers and trainers traded through forums, spreadsheets and phone calls. We built and operate the marketplace that brought them together, with paid listings, built-in payments and ten times easier discovery.
- [Carbon Analytics Web Platform](https://www.altslate.com/portfolio/carbon-analytics): A carbon-monitoring business had to turn 2.4 million land parcels into figures its customers and auditors could trust. We designed, built and now operate the platform: one system of record, with answers in under five seconds.
- [Data & Analytics Platform](https://www.altslate.com/portfolio/joyn-analytics): An upstream oil and gas operator had production, accounting and finance data it could not act on. With JOYN Analytics, teams ask questions in plain English, build their own dashboards and schedule reports, without waiting on IT.
- [Boosting Field Efficiency with Intelligent Task Automation](https://www.altslate.com/portfolio/field-automation): An oil and gas field operator paid twice for reactive maintenance: lost output and crews driving to healthy equipment. Our system watches the field's sensor feeds, flags trouble early and routes the nearest qualified technician.
- [Algorithmic Portfolio Compression for an International Bank](https://www.altslate.com/portfolio/portfolio-compression): A leading international bank's rates desk carried redundant trades that tied up regulatory capital. Our compression algorithms cut the trade count by 20% and notional by EUR 7 billion, with the desk's market-risk position unchanged.

## Articles

- [Half the Memory, None of the Answers Lost](https://www.altslate.com/blog/half-the-kv-cache): Small AI models get expensive when they must remember a long history. One fix flopped. The other halved that memory for free, without changing an answer.
- [The Verifier's Blind Spot](https://www.altslate.com/blog/the-verifiers-blind-spot): We built a "self-improving" LLM cascade to get cheaper every round. It got quietly worse — and every dashboard said it was fine. What we measured, and why you cannot see it from inside.
- [Technology Finally Reached the Delivery Layer](https://www.altslate.com/blog/technology-reached-the-delivery-layer): Cloud rebuilt the front door of services firms and left the workshop untouched. Agents walk into the workshop — and that changes the economics of professional services.
- [Why We Build Agent-First, Not Agent-Bolted-On](https://www.altslate.com/blog/why-agent-first-architecture): Bolting an LLM onto an existing app gets you a demo. Designing the agent as the operating core — with humans in command — gets you a product. Here is how we think about the difference.
- [Serverless That Actually Scales (and Stays Cheap)](https://www.altslate.com/blog/serverless-that-actually-scales): Serverless is not automatically cheap or scalable — it is a set of trade-offs. Here are the patterns we lean on to get fan-out compute that bills to zero when idle.
- [Production, Not Pilots — How We Build](https://www.altslate.com/blog/production-not-pilots): Most AI work stalls at the demo. We design, build, and operate production systems end to end — applying AI where it earns its place and conventional engineering everywhere else.

## Optional

- [Privacy Policy](https://www.altslate.com/privacy): How enquiry details, cookies and browser storage are handled
- [Sitemap](https://www.altslate.com/sitemap.xml): Every page on the site