Skip to content
Services

Productized consulting offers

Each engagement is scoped, deliverable-bound, and designed to produce a measurable outcome — not an open-ended retainer.

01

Conversational Data Agents

Ask anything of your data in plain English — and get an answer you can defend.

We build AI agents that sit on top of your governed data and turn everyday questions into validated SQL, returning a written answer, a downloadable Excel table, and a chart where the question warrants one. Business definitions live in a single versioned source of truth, so the same question returns the same number no matter who asks it. Every query the agent writes is logged, reviewable, and regression-tested against a suite of known-good answers.

Best for: Organizations sitting on large, well-structured datasets where analysts have become the bottleneck for routine questions — and leaders who want self-serve access without giving up control of the numbers.

View live demo

Deliverables

  • Governed semantic layer
  • Natural-language query agent
  • Query guardrails & audit trail
  • Accuracy evaluation suite
02

Analytics Transformation

Move from ad hoc queries to a governed, self-serve analytics platform.

A structured engagement to modernize your data and analytics capability — from data model design and pipeline engineering to dashboard standards and data literacy. Built for organizations ready to make analytics a competitive advantage.

Best for: Organizations with fragmented data, inconsistent reporting, or a growing need for self-serve analytics.

Deliverables

  • Data model & pipeline design
  • Platform architecture
  • Governance framework
  • Team training plan
03

Executive Dashboards & Decision Systems

Executive-grade dashboards your leadership team trusts and uses daily.

Design and build of decision-support dashboards in Power BI, Tableau, or custom web applications. Includes KPI definition workshops, semantic model design, and iteration until the output drives real decisions.

Best for: Leaders who need a single source of truth across commercial, operational, or financial performance.

Deliverables

  • KPI framework
  • Semantic data model
  • Certified dashboard suite
  • User training & documentation
04

Reporting Automation & Data Engineering

Reports that run themselves — accurate, on time, every time.

Design and implementation of automated reporting pipelines, from source system extraction through transformation and delivery. Includes scheduled distribution, exception alerting, and documentation for long-term maintainability.

Best for: Teams spending hours per week on reports that should be automatic.

Read the case study: contractor safety reporting →

Deliverables

  • Pipeline architecture
  • ETL implementation
  • Scheduled delivery setup
  • Runbook & documentation
Proof of work

Built and running

Consulting claims are easy to make. These are shipped systems, not slides. Where one is public, you can open it and test it yourself.

GO Transit Trip Planner

What time is my train — and if I miss it, when is the next one?

That is the only question it answers. No account, no route map, no fare calculator, no saved trips. Enter where you are going and when you need to be there, and it returns the best train with the ones either side of it, so the answer to "what if I miss this" is already on screen.

The restraint is the point, and it is the same discipline that decides whether a dashboard gets used: name the decision the thing supports, then refuse everything that does not serve it.

Underneath, all 42 station codes were matched against Metrolinx’s official GTFS feed rather than guessed, and testing against the live API — not the documentation — surfaced nine defects. One was a timestamp parser reading minutes as the hour: it returned times that were wrong while parsing perfectly cleanly. Nothing crashed. It quietly gave the wrong answer, which is the failure mode that actually matters in reporting.

  • Public API integration
  • Source-verified reference data
  • Live-data validation
  • Containerized deployment
Open the app (opens in a new tab)

Editorial Agent

Is there anything worth saying this week — and would I put my name on it?

Twice a week it reads its source list, scores every candidate against a rubric I wrote, and shortlists at most three. If nothing clears the bar it shortlists nothing, because a weak post is worse than no post.

Nothing publishes without me. The shortlist goes to a private channel for approval, and if I do not answer, the candidate expires and is logged as lapsed. The gate fails closed: silence means nothing goes out, not that something slips through unreviewed.

Every outcome is recorded — published, edited, skipped, lapsed — so the rubric is tuned against what I actually approve rather than what it guessed I would like. The source list got the same treatment: every feed was verified with a real request, and two that returned a clean 200 were dropped because their newest items were from 2020 and 2015. A green status code is not evidence of a live source.

  • Scheduled LLM pipeline
  • Human-in-the-loop approval
  • Decision log & tuning loop
  • Self-hosted orchestration

Not sure which engagement fits? Let's talk through your situation.

Book a discovery call