About AppStream

We bring the AI frontier into the enterprise.

Frontier labs create new capabilities. We work out where they matter in your business, engineer them into real workflows, and keep them reliable in production.

Talk to an AI engineerHow we work
in delivered value across enterprise engagements
$100M+
in delivered value across enterprise engagements
average efficiency gain, measured across client work
285%
average efficiency gain, measured across client work
people using systems we built
5M+
people using systems we built
years building enterprise systems
15+
years building enterprise systems

Why AppStream exists

The frontier moves faster than businesses can absorb it. Somebody has to close the gap.

Every few months AI can do something that wasn’t practical before. Reason over policy. Operate software. Answer the phone. But a model being capable of something and a business benefiting from it are far apart.

Strategy firms stop at the roadmap. Software firms treat AI as one more API. AI boutiques stall when a prototype meets identity, integrations, and legacy systems. We built AppStream to do all of it, from the first question to the system running in production.

Our point of view

The model isn’t the advantage. What you build around it is.

Everyone can buy the same models. The advantage is in the context, integrations, permissions, evaluation, and operations around them, and in the workflows redesigned to use them.

Our discipline

Applied AI engineering. Strategy, engineering, and operations as one practice.

We turn emerging AI capabilities into production systems that perform real work. Advise, Engineer, and Operate run in order, with governance under all three. It’s how we scope, staff, and report every engagement.

Our model

01AdviseKnow where to apply AI.Find where emerging AI capabilities can create measurable business value, and build the roadmap to get there.Explore Advise02EngineerTurn possibility into production.Build the applications, agents, integrations, and data infrastructure that put AI inside real business workflows.Explore Engineer03OperateKeep AI working.Own and continuously improve AI systems after they reach production.Explore Operate
GovernMake AI safe to scale.Across all three

Our heritage

Engineers first. AI-native by design.

  1. 01

    Software engineers first

    We started as a custom software firm building platforms for healthcare, proptech, and fintech companies. Clutch named us a top .NET and React developer.

  2. 02

    Early to language models

    We contributed to Microsoft’s Semantic Kernel, built for its first hackathon, and open-sourced tools that turn unstructured data into typed objects.

  3. 03

    AI in production

    Then we put it to work: a reimbursement engine that reasons over Medicare and Medicaid policy, and an HR assistant that serves every company in a PE roll-up.

  4. 04

    Applied AI engineering

    Today that is the whole practice: advise, engineer, and operate, with governance throughout. Anthropic Select, Microsoft, and AWS partner.

Justin TannenbaumCo-Founder & CEO
“Time and materials pays a firm to take longer. We started AppStream to run the other way: a dedicated team at a flat rate, with AI doing the work that used to fill a timesheet. Now getting faster is how we win.”

Principles

How we decide. Opinions we act on.

Not values on a wall. These settle arguments on real projects, and you should hold us to them.

  • Start with the work, not the model.

    We begin with the workflow and what it costs. The technology is chosen last, once we know what the work needs.

  • Production is the finish line.

    A demo proves something is possible. We count it as done when your people depend on it.

  • Use the best model for the job.

    We don’t resell a model or a cloud. Each task goes to whatever does it well, for the least money.

  • Design for change.

    The frontier moves every few months. Systems we build can swap models and add capabilities without a rebuild.

  • Measure business outcomes, not AI activity.

    Tokens used and prompts written don’t matter. Hours returned, errors avoided, and revenue collected do.

  • Keep humans where judgment matters.

    AI takes the steps. People own the decisions that carry risk, and the system makes those decisions easier to get right.

Team

Humans. The people on the call build the system.

A senior team with no hand-off between the people who scope the work and the people who ship it. Forward-deployed engineers work inside your team when the job calls for it.

  • Justin TannenbaumCo-Founder & CEO
  • Lukasz ChmielewskiCo-Founder & CTO
  • Piotrek SzyperskiHead of Engineering
  • Will GrayHead of Go-to-market
  • Cesar BustamanteFounding Sr. Full Stack Engineer
  • Bartek BienkowskiFounding Sr. DevOps Engineer
  • Takami RodriguezSr. Full Stack Engineer
  • Ilia AndrienkoSr. Front End Engineer
  • Daniel BukalaSr. Product Manager
  • Taylor LarsonSr. Product Manager
  • Lukasz WachnikSr. Full Stack Engineer
  • Mikolaj KaminskiSr. Software Engineer
  • Oleksandr KotenkoSr. Front End Engineer
  • Ernesto QuispeSr. Product Designer
  • Bartek SzyperskiFull Stack Software Engineer
  • Klaus BegnisForward Deployed Engineer
  • Francisco CastilhoForward Deployed Engineer
  • Drake LankfordIntern
  • YouSee open positions

Open source

Built in the open. Extracted from client work.

Libraries we pulled out of real engagements so other teams could use them. Most sit on the .NET and Azure side of what we build, including early work on Semantic Kernel.

Bring us the workflow. We’ll tell you what’s possible.

Tell us which process costs your business the most. An AI engineer will tell you whether today’s AI can change it, and if the answer is no, we’ll say so.

Talk to an AI engineerSee our work

Looking for a job rather than a partner? See open positions.