Skip to content
SepalstackSepalstack
Menu

Software, data and AI studio

Software that grows with you, built in layers.

Sepalstack designs and builds custom software, data platforms and AI agents for teams that need them to work in production, not just in a demo. Each layer ships on its own, so you see value in weeks.

Start a project Explore services
unstructured datasets made searchable by AI agents
25,000+
application platforms delivered
35
working with large, spatial and messy data
14 yrs

Figures describe projects our founder led before and during Sepalstack.

Services

Ten ways we help, one team doing all of it

From the interface people click to the data and cloud underneath, we build every layer, so nothing gets lost between vendors.

Industries

Deep roots in data-heavy industries

We started in energy and geophysics, then built for banks, global brands, security vendors and carriers. The problems rhyme: too much data, too many systems, too little time.

How we work

Small releases, every week, until it is done

No six-month silence followed by a big reveal. You see working software early and steer as we go.

  1. Discover

    1–2 weeks

    Workshops with the people who will use it. You get a written plan, a fixed scope and a price.

  2. Build in layers

    Weekly

    A working release every week on a live staging link. You review, we adjust.

  3. Launch

    1 week

    Production release with monitoring, documentation and a walkthrough for your team.

  4. Support

    Ongoing

    Optional monthly support, cloud management and new layers as you grow.

Selected experience

We have done this at scale before

A few projects our founder led at previous firms, described without naming clients:

  • Banking

    Document AI for a large US bank

    Retrieval and multi-agent systems that read legal and financial documents for lending, risk, investigations and procurement teams, with audit trails and human review built in.

    • AI agents
    • RAG
    • Governance

  • Consumer brands

    Ask-your-data assistants for global brands

    A chatbot that turns plain-English questions into SQL and answers with a chart, alongside semantic search over documents and AI video transcription.

    • Text to SQL
    • Search
    • Google Cloud

  • Energy finance

    Pricing and due-diligence platforms for energy assets

    Seven products built from the ground up, including interactive maps of wells and leases, pricing models and secure, streaming APIs on AWS.

    • Maps
    • Fintech
    • AWS

  • Cloud data platform

    Predictive analytics before launch day

    Pre-launch engineering on in-warehouse machine learning features that were announced to developers worldwide at a major conference in 2024.

    • BigQuery ML
    • Analytics

Tools

Proven tools, chosen for the job

We are not tied to one vendor. We pick what fits your team, your data and your budget, and we write down why.

AI

  • Vertex AI
  • Claude
  • Gemini
  • Agent Development Kit
  • RAG
  • Vector search

Data

  • BigQuery
  • PostgreSQL
  • PostGIS
  • Airflow
  • Dataflow
  • MongoDB

Cloud

  • Google Cloud
  • AWS
  • Terraform
  • Cloud Run
  • Lambda
  • Docker

Interface

  • React
  • TypeScript
  • Next.js
  • D3.js
  • Mapbox
  • ArcGIS

Questions

Before you get in touch

Anything else? Ask us directly.

What does a typical project look like?

Most start with a one- to two-week discovery that ends in a written plan and a fixed price. We then build in weekly releases you can try on a staging link, launch, and hand over with documentation and a walkthrough.

How do you price work?

Discovery is a small fixed fee. After that, most builds are priced as a fixed fee per phase so you know the cost up front. Ongoing support and cloud management are billed monthly.

Who owns the code?

You do. Code lives in your repository and infrastructure runs in your cloud accounts from the start, so you can bring the work in-house at any time.

Is our data safe if we use AI?

We deploy models inside your own cloud project wherever possible, choose providers that do not train on your data, and log every request so answers can be audited.

Can you work alongside our own engineers?

Yes. We often build the first layers, then pair with your team so they can carry the product forward.

Where are you based?

Houston, Texas. We work with clients across the United States, remotely and on site when it helps.

Have a project in mind?

Tell us what you are trying to do. We reply within one business day, and the first call is free.

Start a project