Internal tools and dashboards

An internal tool from Mytrya is a small web application, usually one page and one API route, that pulls data from a system your team already uses, computes the numbers your team asks for, and presents them without anyone running a report.

Where a model is involved, it does the one thing a database query can't: reading two hundred support tickets and naming the five recurring problems. It is never asked to add anything up.

A good fit when

  • A question comes up every week that someone answers by hand: what is support hearing, what is the team working on, which leads went cold.
  • The data is in a system with an API: Freshdesk, monday.com, HubSpot, a Postgres database, a spreadsheet you export.
  • The audience is your own team, so the tool can be plain and fast rather than polished for customers.

Not the right call when

  • You need a customer-facing product with accounts, billing and design polish. That is a product build, and I'd scope it differently.
  • The data doesn't exist in any system yet. Collecting it is the first project.

What gets built

6 parts

  • Scheduled fetches on a cron, with results cached so the first person in doesn't wait
  • Statistics computed in code and tested: counts, medians, resolution times, ageing
  • A model pass for clustering and narrative, bounded to a fixed number of records per run
  • Request coalescing so several people opening the page at 9am trigger one fetch
  • A single page with filters and a table, linking back to the source system
  • A refresh control gated behind a shared secret, and a cost ceiling per day

How it's kept safe

  • Numbers never come from the model. If a figure appears on the page, a function computed it and a test checks it.
  • Every expensive step has a cap that depends on the window being viewed, so a 30-day view costs about what a 24-hour view does.
  • Read-only against your systems by default. Writes are a separate, later decision.

Typical stack

  • Next.js, deployed to Vercel or Cloudflare
  • Redis or Postgres for caching and history
  • Gemini Flash or Claude Haiku for bounded model passes
  • Recharts for the few charts that earn their place

Built and running

2 case studies

  1. Support IntelligenceA morning read of a support desk for people who don't read ticketsNext.js · Gemini 2.5 Flash · RedisIn productionClient work
  2. Community SignalTwo small internal tools that each replaced a weekly manual taskPython · SQLite · Claude / Next.js · monday.com GraphQL · GeminiInternalClient work

Questions about this service

How long does an internal dashboard take to build?

The two on this site were each a few days of work against an existing API. The build is small. Most of the time goes into agreeing which questions the page must answer and how the numbers are defined, and I do that in writing before building.

Can the tool write back to our systems, not just read?

Yes, but I build the read-only version first and run it for a while. Writes come as a second step with their own dry-run mode, once the team trusts the numbers.

Next step

Have a process that runs on someone copying data between two tabs?

Describe it: what it is, who does it, how often, and what goes wrong when it's late. I reply within one working day with a scoping call or a reason it isn't worth automating.

Start a project How a project runs

Free 30-minute call · one working day