About
A specialist practice, not a consultancy pyramid
Info Insight Limited is a UK analytics consultancy working with local authorities and housing associations on predictive models and simulations. Work is delivered personally by the Managing Director, inside the client's own environment, and handed over in full.
The practice
Built around one question: what happens next, and what will it cost?
Public sector organisations are asked to set annual budgets against demand that is generated by thousands of individual cases, each with its own trajectory. The traditional response is to uplift last year's outturn and hope. Info Insight exists to replace that with something defensible.
The methods are conventional and well understood — supervised machine learning, stochastic and discrete‑event simulation, economic evaluation — applied carefully to messy administrative data by someone who understands what the data records and what it does not.
Engagements are typically fixed‑price and scoped around a single product with a clear deliverable, rather than an open‑ended advisory retainer.
Ryszard Sekula
Managing Director
All work is delivered personally, over secure encrypted connections, in full compliance with applicable data protection legislation. Info Insight does not use sub‑contractors.
How we work
Four stages, and you own everything at the end
Scope and data
We agree the question, the decision it supports and the minimum data required. Source data is pseudonymised and record‑level, typically covering five or more financial years.
Build in your tenancy
Development happens in a client‑provisioned workspace with Git version control and experiment tracking. Multiple algorithms are trained, validated and competed before one is selected.
Validate and hand over
Documented notebooks, written methodology, a retraining runbook and a walkthrough session with your analysts covering the pipeline end to end.
Refresh and support
Optional quarterly retrains on the latest completed period. You can run these yourself, or have each retrain delivered and validated as a managed service.
Typical environment requirements
- A dedicated workspace, with Info Insight added as a guest member or contributor via your identity provider
- A lakehouse or warehouse to hold the pseudonymised source data
- A notebook environment permitting standard Python data science packages — LightGBM, XGBoost, scikit‑learn, pandas, numpy
- A pipeline or dataflow for ingesting and refreshing source extracts
- Git integration for version control, and experiment tracking for the model registry
Compute demand is modest. No specialist capacity is required.
Data and security
No record‑level data leaves your organisation
Models are built entirely within the client's own cloud environment — typically Microsoft Fabric, or an equivalent platform. Info Insight works inside a workspace you provision and control, which keeps you in full control of sensitive record‑level data and meets your data protection obligations.
Source data should be pseudonymised before it reaches the workspace, with identifiers applied consistently across years and refreshes so that the longitudinal patterns the models rely on are preserved.
All code and model artefacts are held in your own Git repository. Nothing is dependent on Info Insight after handover.
Info Insight is happy to provide whatever documentation your procurement, information governance or security teams require to support approval and onboarding.
Handover and training
The point is that you can run it without us
Documentation
Commented notebooks and a written overview of the data flow, feature definitions and model configuration — plus a retraining runbook setting out how to refresh the model on a new period of data.
Training and support
Hands‑on training in the machine learning models and process simulation methods used, and a walkthrough session with your analysts covering the pipeline end to end, to ensure effective handover and continued use.
Technology transfer
Machine learning and predictive analytics methods are shared with the organisation rather than held back, including plain‑language descriptions of the advanced analytics used, for client transparency and audit.
Commercial
Fixed price, fixed scope, clear timeline
- Fixed‑price packages. Each product is quoted as a single fixed price, excluding VAT, with optional extras priced separately and commissioned only if wanted.
- Typical build timeline. Three months from receipt of data and workspace access for a first model, with quarterly updates following once the relevant period's data is complete.
- Payment terms. Part‑payment on commencement following receipt of a purchase order, with the balance on final delivery.
- Procurement support. Whatever documentation your procurement team needs to support approval and onboarding.
Start with one product and one dataset
Most engagements begin with a single model, built to answer a single question that is currently hard to defend. Tell us which one it is.