Layered Studio
Better access to what your organisation already knows, and better tools to plan the next programme.
Organisations hold years of evidence in evaluations, reviews and monitoring data, and most of it is hard to use once it is filed. We make it easy to find, read and trust, and we build tools that carry its lessons into the design of the next programme. Evaluation synthesis, from a single evaluation to a meta-synthesis across many, is the first strand.
Our approach
The same flow at every scale: what a programme collects becomes structured records, records become an evaluation, and many evaluations are read together as one.
How AI is used
We use AI where it is good, reading a great deal of text, and keep it away from what has to be exact. So every finding can be traced to the page it came from.
Model
A model reads every page of every report and pulls out its findings and recommendations, each with the page it sits on. For a briefing, it puts the evidence in order for the reader and drafts the framing.
It never writes a citation or a number.
Code
Every quotation is looked for on the page it cites, and one that cannot be found there is never shown. Every count, date and place on a briefing is worked out from the records themselves.
Citations are built from the source.
Person
Evaluators judge each evaluation's findings. Every citation opens at its page, so a person can judge what code cannot: whether the quote supports the claim, and whether it is accurate and fair to its context.
Judgement stays with people.
Our engine
Whoever asks gets a report written for them, for their role and their question, drawn from all of a project's documents and data, spatial data included, as text, graphics and charts, every point cited to its source.
We build it to be adapted to each project, urban and humanitarian alike.
Our engine
The engine scores a new proposal or plan against what earlier work found, criterion by criterion.
Each score comes with cited strengths, gaps and suggested changes.
Contact