GSMA is already tendering the endline review of WonderTree's grant, with 40 interviews due by April
The funder's own paperwork schedules the endline research on the therapy games company, with bids for the fieldwork due on 23 November.
WonderTree, the Pakistani company that turns therapy exercises into motion games for children with special needs, is heading into an evaluation by its funder. The GSMA, which runs the UK funded Innovation Fund for Impactful AI, has published terms of reference for a Pakistan based consultant to carry out 40 in person interviews in March and April 2027 as the endline assessment of WonderTree's grant. Bids close on 23 November.
The terms of reference are the primary record. They describe WonderTree as a company that uses AI and augmented reality to turn physical therapy and special education exercises into games for children with mild to moderate motor and cognitive challenges. The GSMA says it will lead the research design, sampling, interview guides, analysis and final report. The hired consultant handles fieldwork: recruiting respondents with WonderTree's help, interviewing them in person, and delivering recordings, verbatim transcripts and English translations.
Who gets interviewed
The 40 interviews are indicative and may change once the GSMA checks who can be reached. The document names teachers, school administrators, parents and guardians whose children use WonderTree's technology as the likely respondents. It asks the consultant to explore changes in early childhood development, including motor skills, and how the effect differs across use cases and devices, and it asks for particular care when parents discuss sensitive health experiences.
The schedule is tight. Contracting and an inception meeting are set for December 2026, adjustments to the interview guides in January 2027, recruitment in February, fieldwork in March and April, and handover of materials in May. Forty interviews across about two months is roughly five a week, which is a manageable pace for one team but leaves little room if schools are closed or families cannot be reached. The GSMA does the analysis after the handover.
How bids are scored
The GSMA says it cannot give a budget at this stage, so bidders must price blind. The criteria weigh cost at 20 percent, quality of method at 35 percent, capacity to deliver on time and on budget at 30 percent, and relevant experience at 15 percent. Cost is therefore the smallest lever: 80 percent of the score rests on whether a bidder can show it can reach families and finish on schedule. The terms ask for qualitative fieldwork experience in the education technology sector in Pakistan, research in early childhood development and assistive technology, and transcription in local languages with accurate English translation.
The dates imply a late 2025 start
The terms of reference do not give the grant's start date or size. The fund's own terms for the Impactful AI round, published in January 2025, say grants of £100,000 to £250,000 last 15 to 18 months and require the applicant to put in at least 25 percent of the grant as matching funds. If the endline falls near the end of the grant, counting back from fieldwork in March and April 2027 puts the start between about September 2025 and January 2026. That is an inference from the fund's rules, not a figure the GSMA has published for WonderTree, and the endline could come before the grant ends.
TechJuice reports that WonderTree is one of eight grantees chosen from 625 applications across 40 countries, and that it plans a mobile app to make its platform cheaper and easier to reach, along with early childhood development activities covering cognitive, motor and literacy skills.
What the evaluation can and cannot show
The design of this endline is sensible for a small grant. Interviews with teachers and parents can show whether children use the games regularly, whether adults think motor skills have improved, and where the devices or the software get in the way. They cannot show how large the effect is. Forty interviews, recruited with help from the company being evaluated, are a sample of experiences and not a measurement, and the terms of reference do not mention a comparison group or any clinical test. Any claim of improvement that comes out of the report should be read as reported by adults who chose to take part.
The mobile app is where the economics will be decided. UNICEF describes the games as using human pose estimation, which means software reading a child's body movements from a camera. That works on a school tablet or laptop with a steady view of the child, and is harder on the shared, low cost phones many families use, where a slow processor or a weak camera limits what the software can detect. A cheaper app would widen the market, but none of the sources gives figures on how many children use the product or what schools pay.
WonderTree is not new to outside funders. In 2020 UNICEF's Innovation Fund selected it as the only Pakistani startup among 12 chosen from more than 350 submissions from over 60 countries. The next dates are 23 November, when bids for the fieldwork contract close, and March 2027, when the interviews begin.
