USAID / BHA
A customized language-model workflow for labeling and interpreting humanitarian strategies.
Project overview
USAID/BHA's portfolio materials contained strategic approaches that were difficult to classify consistently by hand. A customized language-model workflow detects labels such as Refine and Implement, Theory of Change, and Collective Impact, then pairs the results with visualization tools so teams can review strategy patterns faster.
Project story
USAID/BHA needed to interpret strategic approaches across a large body of humanitarian planning material. Approaches such as Refine and Implement, Theory of Change, and Collective Impact were important to detect, but manual review made it difficult to label them consistently and keep the evidence base current.
The customized language-model workflow improved classification accuracy and paired the labels with visualization tools. That made findings easier to communicate and review. The work saved recurring staff time, increased the amount of evidence available for decisions, and helped teams move faster through portfolio interpretation without losing expert validation.
Example deliverables
Concrete outputs or working assets the project produced or supported.
Decision problem
USAID/BHA needed a way to classify and interpret strategic approaches across a large body of humanitarian planning and portfolio material. Teams were looking for recurring approaches such as Refine and Implement, Theory of Change, and Collective Impact, but manual review made it difficult to keep the evidence base current.
What the work involved
Customized a language-model workflow to detect and label strategic approaches in BHA materials, improving consistency in how planning concepts were identified.
Paired the classification workflow with visualization tools so teams could interpret findings faster and communicate patterns more clearly.
Created a continuously updated evidence base for portfolio decision-making, reducing manual analysis time and improving evidence use across humanitarian strategy review.
Outcome
30% increase in labeling accuracy, 98 staff hours saved per month, and stronger evidence use through improved visualization and analytics.
Evidence inputs
Review focus
Bring one decision, report, review process, or evidence problem. We will help identify the smallest useful place to begin.
We use cookies to enhance your browsing experience on our website. By continuing to use our site, you consent to our use of cookies. For more information, please review our Privacy Policy.
