Before an agent
What the CRM, inbox or file store actually holds.
Data Science Consulting
Write down the data before you add a model
Most AI work fails on the records, not on the model. We write down what you have, what is missing and what a person must still see.
- The output is a written data map and a first job, not a dashboard of invented metrics.
- We stay with the products and agents we know how to ship.
- We do not publish research papers or sell a data science bench.
Data mapFirst jobHuman review
What the review covers
What you already store
The fields, files and logs the product already writes.
What a model may read
The slice that is enough for the first job, and no more.
Who reviews the output
The person who can reject a draft, a check or a booking.
Choose a starting point
A process review is the usual first step. You can also look at related work or bring a specific workflow.
What you leave with
A data map
Sources, owners and the gaps that would block a build.
A first job
One task an agent or a model could take, if the data is enough.
What the system should not read
Fields the system should not read, and steps that stay with a person.
A build if you want one
The same team can then put that job in the product.
Where the review sits
Before a care feature
What a clinic record contains and who may see it.
Before a check or a score
Whether you have labels a reviewer can agree on.
A decision to wait
If the data is not there, the review says so.
Tools we build with
The stack behind the systems on this site.
- Next.js
- React
- React Native
- Node.js
- Python
- PostgreSQL
- AWS
- OpenAI
- Vapi
- Twilio
- n8n
- Make
- Zapier
- GoHighLevel
How the work runs
Process review
We sit with how your team works today and pick the workflow that is worth building first.
Design and scope
You get the shape of the product, the systems it touches, and a scope you can agree.
Build and test
We build, test with your people, and keep a human in the loop where the agent decides.
Launch and support
We launch with you and stay for support after the system is in use.
Brown Tech in numbers
Since 2011
in business
1,000+
apps delivered
99%
client satisfaction
10+
billion-dollar companies served
Who we are
Brown Tech has been in business since 2011. We design and build custom software, AI agents and automation for teams that want software built around how they already work.
We start with the system your team needs, connect it to the tools you already use, and add AI where it removes real work.
Why teams choose Brown Tech
- in business
- Since 2011
- apps delivered
- 1,000+
- client satisfaction
- 99%
- billion-dollar companies served
- 10+
Other AI work
- ML ServicesModels inside a product we know how to ship.
- AI ConsultingFind what to build first, what it will cost and what can wait.
- Big Data ServicesData stores and APIs under the products we ship.
- AI Software DevelopmentSoftware with AI in the workflow your team already runs.
- HIPAA Readiness ReviewA review of access, logging and review in a care system.
- Custom AI AgentsAgents that handle real tasks in the tools you already run.
Questions about this work
No. It is a review of the data a product or an agent would use, written so a build can start or wait.
No. A model in a product is a separate scope, on the ML services page.
Timing is agreed in the process review, project by project. That review comes before build starts.
Cost is scoped per project after the process review.
Support after launch is part of how we work. The shape of that support is written into the scope.
No. We only write down figures Brown Tech already publishes, or numbers that come from your own records.