SAAS FOR MONITORING AND EVALUATION
The report is written by AI, from numbers a person confirmed.
SpectoQ is a platform for the monitoring and evaluation of donor programmes and strategies. An organisation builds the structure of a project in it, with its own levels and indicators, the people responsible follow and measure those indicators, and then somebody has to verify the results. The crown of the system is its reporting, thorough and flexible, with narratives written with the help of artificial intelligence. We built the platform for Profectus, and we are still improving it.
We understood the work before we wrote the code.
Monitoring and evaluation tools existed, but they covered only part of the job, with no AI, limited flexibility, rigid roles and limited support. The organisation had to bend to the software instead of the other way round.
So the first step was not code. Together with the client’s M&E team we went through their business processes and through what the existing systems could not do. We began with the questions at the core, what is measured, how it is measured, who verifies it and who the report goes to. Out of that analysis came the structure and the design that still hold the product up.
One structure, three lives.
All of it from scratch. From paper on a whiteboard, through a diagram, to a live screen.
Structure and levels Young people in the labour market 2025-2027 · in progress Impact 3 ind. Less youth unemployment across 12 municipalities 67.5% 1. Outcome 2 ind. Young people find work faster after training 33.33% 1.1. Output 4 ind. 400 young people trained for jobs in demand 50% 1.1.1. Activity 2 ind. Workshops in 12 municipalities 24% From an indicator to an AI report.
01 / 05 A framework per donor
02 / 05 A measure per result
03 / 05 Evidence, verified
04 / 05 A locked target
★ / 05 A report in minutes
Donor projects, logframe programmes and city strategies stand side by side, each on its type.
A type carries its own chain of levels, so a donor ToC and a city strategy live in one system.
An intervention is built along that chain, and every level carries its own indicators and its own percentage.
Baseline 50,000, target 70,000, reached 72,000, over six months with a dip in December.
The agreement is signed, the first interns wait on verification, the contract offer is ahead.
Three links feed the sum, one from another project entirely: $185,000 of a $250,000 target.
A value, an explanation and an attached file, because without the evidence nothing moves.
The verifier sees who measured and on what basis, then accepts it or sends it back.
When the audit comes, the system already holds who measured, who confirmed it, when, and on what note.
Before the start the framework is built freely, and the app warns what closes after it.
A target is not changed in a field but through a request, with an explanation that travels.
You see who is asking, why, what changes and what stays. The administrator decides.
One field, then Generate. What goes into the report the system already knows from the chain.
Indicator by indicator, 2 of 33, through collection, processing and assembly of the document.
The tables arrive filled in, and a slash command drops in the intervention, its indicators or the measurement images.
The screens come from a development build of the application, filled in through a scenario. The rules, the flows and the controls are real; the numbers are not data from real programmes.4 The AI report screens are cleaned-up builds, and no generated prose appears on them.5
Software people recommend.
The words come from people who work in the system every day, at an institute, a network of organisations and a municipal administration.6
The system is efficient, adaptable and reliable, and it saves time. It has improved our reporting and the way we make informed decisions.
Aziz Đipa Head of M&E · Institut KULT For years we tried many similar tools, but we never found one for monitoring and evaluation flexible enough for the nature of our work.
Mahir Poljić Project manager · CCI For a long time we struggled to collect data on the progress of indicators and to tie it clearly to our strategic goals. We used several tools for reporting and tracking, and that slowed the whole process down.
Lejla Dizdarević Assistant to the mayor · Centar Municipality A SaaS, not a project.
SpectoQ is sold from a public price list and today carries the programmes of three organisations. Subscriptions, terms of use and user support run through a back office we built. The client runs a business, not a codebase.
TECHNOLOGY
SOURCES Where everything on this page comes from
1 WHO USES IT
Institut KULT, Centri civilnih inicijativa and the Centar Sarajevo municipality are users who have said so publicly themselves, on spectoq.com.
2 THE PHOTOGRAPH
Our own archive, a workshop with the M&E team, 11 November 2021.
3 THE PRODUCT PANEL
The third frame in the strip carries an illustrative plan. A click opens the original screenshot.
4 THE SCREENS
A development build of the app, filled in through a scenario. The rules, the flows and the controls are real; the numbers are not data from real programmes.
5 THE AI REPORT
The AI report screens are cleaned-up builds, the same controls and the same data without the surrounding furniture. No generated prose appears on this page; what is shown is the flow, not text the model wrote.
6 THE QUOTES
The quotes come from spectoq.com, where the users gave them publicly. The English wording is our translation of the Bosnian original.
WHAT WE DO NOT CLAIM
We do not claim the programmes’ results as ours, because the platform measures and documents what the organisations themselves achieve and verify.
Do you run a programme that has to prove itself?
We build platforms that answer to auditors. And we stay on to operate them.
