AI-supported behavioral therapy
Developed agentic intensive behavioral therapy for Medicare patients with Bonus Health.
Healthcare
Custom software for your operations.Built. Tested. Delivered.
Custom applications, AI agents and integrations from a senior team across software, data and machine learning.
Teams we have built for
Selected work
Software built around real operations.
Selected engagements across healthcare, biotech and private markets. See what Lokos contributed to each project.
Swipe to see more
Machine learning for biotech
Built a platform and machine learning algorithms to analyze cannabis product composition.
Biotech
Private market intelligence
Led development of systems that extract document engagement signals and use machine learning to analyze private market activity.
Private Markets
The Lokos approach
Useful software, without unnecessary overhead.
Enhance your team.
Automate processes and give your team more room to apply its domain expertise. We work alongside you as a trusted software and AI partner, building the tools that help your business grow.
Start seeing value sooner.
Begin with a useful first deliverable, put it in your team's hands, and improve it in focused iterations rather than waiting for one enormous launch.
Transparency you can rely on.
Clear scope, agreed pricing, and regular progress updates keep you informed. Our engineers review and test the work against agreed acceptance criteria, so you can rely on what we deliver.
The firm
A senior AI engineering team that ships production systems.
Lokos AI builds custom applications, AI agents and business automations. Our team combines software engineering, data and machine learning expertise to turn operational needs into working systems.
Our delivery model combines AI development tools, reusable engineering tools and a structured process for scoping, reviewing, testing and delivering software. Engineers remain responsible for technical decisions and the finished system.
Start with a defined project and an agreed price. Return with additional requests as your needs evolve, supported by a client portal for proposals, payments and delivery updates.
We work with your team to define the requirements, build the software and deploy it into your existing systems.
A small, multidisciplinary team
Machine learning, software engineering and product experience in one team.
Embedded delivery
We work directly with the people who will use the software.
End-to-end ownership
Research, data, models, product, and deployment.
Built for production
Code review, automated tests and checks before release.
What we build
What should your software handle next?
Examples of what we can build around your needs, from automating a workflow to developing a custom application or AI assistant.
01ApplicationsCustom applications and workflow automation
Build the internal tools, client portals and automated workflows your team needs, so information moves between systems with fewer manual handoffs.
- Example deliverable
- An operations portal that collects documents, extracts information and routes exceptions for review.
- A possible first project
- One core workflow, defined user roles and the screens needed to support it.
Common problems Work spread across spreadsheets, disconnected tools and manual handoffs.
02AI assistanceAI agents and document workflows
Build assistants that answer questions from approved sources, extract information from documents and support follow-up tasks, with review and escalation rules defined for the workflow.
- Example deliverable
- An AI assistant that prepares information for a person to verify.
- A possible first project
- One bounded assistant task, agreed information sources and human review criteria.
Common problems Routine questions, document review and follow-ups that consume staff time.
03Connected systemsIntegrations and ongoing development
Connect your existing tools and extend your software as your business changes, through additional projects or an agreed ongoing engagement.
- Example deliverable
- A workflow that keeps your CRM and internal tools synchronized.
- A possible first project
- One integration, agreed field mappings and checks for synchronization errors.
Common problems Duplicate data entry and systems that do not share information.
How it works
From your first request to your next improvement.
Use your client portal to submit requests, review proposals, pay invoices and follow progress. Our team handles scoping, engineering and delivery.
Put the working agreement in placeWe arrange a master services agreement for the relationship. Each project has its own statement of work.
Describe what you need
Define the first deliverable
Review and sign your proposal
Pay your invoice
Follow delivery
Request the next improvementReturn with another request as your needs evolve. We agree the scope, price and delivery plan before additional work begins.
Start with a clear scope and a fixed project price.
Why Lokos
An AI software factory, led by engineers.
We combine AI coding tools, reusable engineering tools and a structured delivery process. Projects move through scope, approval, implementation, testing and delivery, with senior engineers responsible for the result. You work directly with the team building your software and can return with the next improvement as your needs evolve.
- Software, data and machine learning expertise in one team.
- Defined deliverables and acceptance criteria before development.
- Engineer-reviewed code, testing and deployment.
About
About Us.
- Trained at
- NYU Center for Data Science
- Previously
- Podz → Spotify
- Countries
- VE, ES, UK, IL, CN, US
- in/dlcasabuena
The founder
Diego Casabuena founded Lokos after leading machine learning at Podz and building AI products at Spotify. A published researcher and holder of granted U.S. patents, he brings experience across AI research, early-stage businesses, and large-scale production systems.
At Podz, Diego developed fine-tuned podcast transcription models that powered its audio discovery feed before Spotify acquired the company. Over three years at Spotify, he progressed to Senior Machine Learning Engineer. He shipped podcast previews and AI voice translation for podcasters including Lex Fridman, and contributed to Spotify DJ and Automix.
Diego earned his bachelor’s degree from NYU, majoring in Computer Science and Economics, and was a Phi Beta Kappa member and Presidential Honors Scholar. He conducted honors research under Rob Fergus and trained as a researcher at the NYU Center for Data Science.
His technical work spans language modeling, speaker diarization, clustering, drone calibration, and large-scale machine learning engineering. He also co-founded technology companies in the United States and Spain.
How we work
Start with a specific business problem.
We look for tasks that consume time, introduce errors or limit what your team can do. Together, we define what a useful result would look like.
Build a first version. Improve it with feedback.
We deliver in stages so your team can try the software and help decide what to improve next.
Work directly with the engineering team.
We explain the technical decisions, document the system and help your team use and maintain it.
Questions, answered
A clear path to your first project.
Can we start with one small project?
Yes. We often recommend a focused first deliverable, then expanding after your team has reviewed the result. We'll help define a sensible starting point.
How do you estimate the price?
We review what the software needs to do, its integrations, risks, and acceptance criteria. Then we propose a defined scope, price, and estimated timeline for your approval.
What happens if our requirements change?
We document changes and provide an updated or additional scope and price before beginning work outside the original agreement.
How do you use AI to build our software?
We use AI coding tools to support implementation, testing and investigation. Our engineers define the architecture, review the code and verify the system against the agreed requirements. AI also supports parts of our scoping and delivery process.
Can you work with existing software and data?
Often, yes. We can evaluate integrations with your current tools as part of scoping. Feasibility and any third-party costs are confirmed before they are included in a proposal.
How do payments work?
Payment milestones and any upfront payment are specified in each statement of work. Clients can review agreements and manage payments through the portal.
Fixed-fee or hourly?
Both. We offer fixed-fee projects and hourly engagements. Initial projects have a defined scope, fixed fee, and acceptance criteria agreed before work begins. For ongoing work, we agree on the model that best fits your needs.
Do you offer ongoing development?
Yes. You can start additional projects as your needs evolve. For clients with regular development needs, we can also discuss a recurring development arrangement with clearly defined scope and capacity.
Contact us
What is slowing your team down?
Tell us about a manual workflow, disconnected system or application you need. We'll discuss a practical first project and what we need to scope it.
Start with a conversation. Receive a proposal after we understand the work.
- diego@lokos.ai
- @lokos.ai
- 19 Dutch St, New York, NY 10038
