We build systems that give your business its time back.
Automation and applied AI that lift the manual work off your team.
Fixed scope. Fixed price. Documented and supported.
Fixed-price services
Applied AI for internal systems
We build machine-learning and AI systems that run inside your business — used day to day by engineering, finance, operations, R&D, and support teams. Practical systems that earn their keep: time saved, decisions improved, work automated.
Typical work includes:
- Training and deploying task-specific ML models — classification, detection, and anomaly flagging
- Curating, labeling, and validating real-world (and often messy) datasets
- Building RAG chatbots and assistants that answer over your own documents and data
- Integrating models into the internal tools and workflows your teams already use
- Building lightweight frontends and APIs so the results are actually usable
Ask something we built
This is a working AI agent we built and run on AWS Bedrock in our own account — the same architecture we deploy inside a client's. It isn't a scripted demo: it decides what it needs, writes its own SQL, runs it against a live database, and answers from the result. Every query is read-only and scoped to the data it's allowed to see.
- Which production line had the most downtime?
- What's the defect rate by product?
- What can PJA build for us?
The manufacturing data is synthetic, but nothing else is: the agent really does write the SQL, really does run it, and really does read the answer back. Built with AWS Bedrock and Athena. The same pattern runs inside a client's account, on their data, under their controls — which is the point: you can watch it work before deciding whether to hire us.
Analysis and automation
We build the data pipelines, dashboards, and automations that run inside your business — used day to day by engineering, operations, finance, and management teams. Practical systems that quietly do the work: manual steps removed, data made usable, problems caught early.
Typical work includes:
- Automating the manual workflows, tracking, and handoffs that eat your team’s time
- Building data pipelines that turn raw, messy operational data into clean, usable outputs
- Designing internal dashboards and reporting for engineering and management
- Connecting the tools and systems you already use so data moves without copy-paste
- Surfacing the bottlenecks, anomalies, and inefficiencies hiding in your data
Where the manual work goes
The shape most of our automation work takes: something arrives, it gets checked and cleaned, then it branches to whatever should happen next. Hover or tap a step to see what it does.
- Re-keying the same export every Monday
- A dashboard that's only current when asked
- Errors caught weeks late
The tools change per client — Power Automate, Python, SQL Server, Power BI — but the shape holds. What you get is the running workflow plus the documentation to change it yourself: fixed scope, fixed price, and a handoff that leaves you in control.
What we work in
We build on the platforms you already run, rather than asking you to move.
Cloud & platforms
- AWS
- Microsoft Azure
- Microsoft 365
- SharePoint
Automation
- Power Automate
- Power Apps
- Python
- Scheduled jobs & scripts
Data & reporting
- Power BI
- SQL Server
- Dataverse
- Excel
Applied AI
- AWS Bedrock
- OpenAI & Anthropic APIs
- Retrieval (RAG)
- Document extraction
Integration
- REST APIs & webhooks
- ERP & MES systems
- LIMS & instrument data
- CSV & flat-file feeds
Running something that isn't listed? That's normal — most of this work is connecting systems that were never meant to talk to each other. Tell us what you have on the call.
Fixed scope. Fixed price. Named exclusions.
Every engagement starts with a 20-minute call. The price in the scope document is the price you pay, and each offer below names what it does not include — because ambiguous scope is what makes fixed-fee work go wrong.
Operations & Data Audit
Fixed price · 1–2 weeks
Two weeks looking at how work actually moves through your business: where data gets re-keyed, where handoffs stall, where reports get rebuilt by hand, and which of it is worth automating first.
You get
- A written findings report — what's breaking, where, and what it costs you in hours
- A prioritized roadmap: quick wins, bigger builds, and what to leave alone
- Effort and cost estimates for each item, so you can decide what to fund
- A 45-minute walkthrough call
You keep the report whether or not you hire us for anything else.
Not included: building anything, software licenses, access remediation, ongoing support.
Workflow Automation Build
Scoped and quoted in writing · 2–4 weeks
One workflow, end to end, in production. The manual tracking, the copy-paste between systems, the report somebody rebuilds every Monday.
You get
- A working automation, running in production
- Documentation and a handoff walkthrough, so your team can run and change it themselves
Built on: Power Automate, Python, or a workflow platform — whichever fits what you already run.
Not included: anything outside the one workflow named in the scope document. Additions go through a written change order.
Operational Dashboard Build
Scoped and quoted in writing · 3–4 weeks
Reporting that reflects reality instead of last month's export. Connected to source systems so it stays current without anyone assembling it.
You get
- A live dashboard and the pipeline feeding it
- Documentation and a training session
Built on: Power BI, or a web dashboard if your data doesn't live in the Microsoft stack.
Not included: data cleanup beyond what the named sources require; new source-system integrations not in scope.
Internal Application Build
Scoped after an audit
When the spreadsheet needs to become software — an app people log into, with permissions, history, and an export that matches whatever your accountant, auditor, or regulator already reads.
We don't quote these blind. The scope comes out of an audit, so the number you get is one we can stand behind.
Ongoing Support
Fixed monthly rate, after a build
Monitoring, break-fix, credential and dependency upkeep, and someone to call when something breaks. Coverage level, response time, and term are set in the proposal alongside the build — quoted as its own line, never folded into the build price.
Not included: new builds, and changes beyond the hours in your coverage level — both are quoted separately.
What happens after the call
-
1
A 20-minute call
You describe what's broken. We tell you whether it's worth an audit or whether you should just go fix it yourself. Some of these are 20-minute problems.
-
2
A one-page scope document
Deliverables, timeline, revision limit, and an explicit list of what is not included. You sign that — not an open-ended agreement.
-
3
The work
You get progress updates, not silence.
-
4
Handoff
Documentation and a walkthrough. You are never locked in — keeping it running afterwards is a service you can buy from us, not a condition of it working.
The price in the scope document is the price you pay.
If we underestimate the work, that's our problem, not yours. Scope changes only happen through a written change order that you approve first — so there's no version of this where the invoice surprises you.
How a project runs
Four fixed-scope steps, ending in working software your team owns and can run.
You already know which spreadsheet is the problem.
The one three people maintain by hand, that breaks every month, that somebody rebuilds from memory when they leave. We replace those with something that runs itself — and we tell you what it costs before we start. No retainer required to find out.
- Repetitive manual reporting or data re-keying between systems
- Disconnected systems with manual handoffs between teams
- Decisions being made on stale data
- Growth outpacing your reporting or admin capacity
- Regulated or technical workflows that need traceability
- “We want AI” with no specific process behind it. We'll tell you that on the call rather than sell you an audit.
- Problems that are really staffing or process problems. Software won't fix those, and we'd rather say so early.
Tell us what's broken.
Describe the system, workflow, or problem and we'll reply within one business day — and tell you whether it's worth paying us to look at it. If it's a fit, the next step is a 20-minute call. Not ready yet? Tick the updates box on the form and we'll stay in touch.
Get in touch