From Manual Financial Processes to AI-Powered Discovery: 31 Findings Across Two Teams in 3 Months
How PedidosYa used Horizon to surface 31 operational insights across Rider Payments and Partner Payments, without blocking a single calendar slot, and built the foundation for a multi-area improvement program across Finance.
Insights Generated
Employees Interviewed
Finance Areas in Discovery
The Challenge
PedidosYa is Latin America's leading food delivery and quick commerce platform, connecting millions of users, businesses, and riders across 15 local markets. With hundreds of analysts, finance specialists, and operations teams coordinating complex, high-volume payment processes daily, PedidosYa needed a way to surface operational inefficiencies at scale, without months of manual interviews.
Rider wallet adjustments and cash tool failures required 3–4 hours/week of manual rework affecting ~2,400 riders weekly, caused by recurring retries in a 40-minute window with no automated resolution. Manual EDC–SAP reconciliation in Bolivia generated 1,000+ discrepancies per week, with ~2 hours of weekly analysis vs. ~40 minutes in other countries, driven by a visualization error requiring line-by-line Excel comparison.
Payout reports and consolidations were extracted manually from BigQuery, ~30 hours/month of preparation across ~800 weekly payments with no automated pipeline. Partner billing control required 4–5 hours/week to cross-reference SAP, Query, and Looker data in Excel (VLOOKUP) and verify in BackOffice, covering 25,000 partners with ~150 weekly discrepancies. Payment report downloads from Banesco consumed ~40 hours/month (~30h reports + ~10h extra due to Kinpos tip download failures).
There was no centralized visibility into where time was being lost across Finance or which workflows had the highest automation potential. Traditional discovery methods would have required months of scheduled interviews to cover even a fraction of the workforce across 15 markets.
The opportunity: automate rider wallet retry logic and reconciliation to eliminate 3–4 h/week of manual rework at scale across markets; replace line-by-line Excel comparison in Bolivia with automated EDC–SAP matching, targeting parity with other countries (~40 min/week); automate BigQuery extraction and payout report generation to recover ~30 h/month across Finance; standardize partner billing control with automated cross-referencing across SAP, Query, and Looker, eliminating manual VLOOKUP workflows for 25,000 partners; resolve Banesco download failures to recover ~40 h/month in payment report preparation; and build a scalable discovery model to extend AI-powered process mapping to additional Finance areas (Tax, Collections, PQP, Rider Care) without adding headcount.
Methodology
The Horizon Approach
Horizon deployed its AI-powered discovery platform at PedidosYa starting with a focused pilot across Rider Payments and Partner Payments, conducting 14+ asynchronous AI interviews in 3 months without blocking a single calendar slot.
The platform's conversational AI adapted in real time, surfacing operational patterns and quantifying time loss across workflows that manual interviews consistently miss. Based on the pilot results, PedidosYa expanded Horizon's deployment to additional Finance areas including Tax, Collections, and cross-market process benchmarking (PQP).
Results: 31 actionable findings generated across two financial teams in the pilot. Quantified waste across 5 high-impact process areas, from rider reconciliation to partner billing. Hundreds of hours per month of operational overhead surfaced and documented. Process documentation generated and structured for internal knowledge management and junior team onboarding. Discovery extended to Tax, Collections, and PQP, building a continuous, scalable improvement pipeline across Finance. Employee feedback confirmed high platform adoption: described as intuitive, rational, and accurate in voice transcription.
Traditional consulting would have required months of scheduled interviews and manual analysis across 15 markets. Horizon delivered 31 findings across two teams in a single quarter, asynchronously, and scaled to additional areas without restarting the process.
The Results
The PedidosYa engagement delivered more than a process audit. Horizon gave the Finance team a clear, quantified picture of where operational time was being lost, and built the infrastructure to keep discovering it.
Starting with a focused pilot across Rider Payments and Partner Payments, the team surfaced 31 findings, documented hundreds of hours of monthly waste, and established the foundation for a multi-area improvement program across Finance in 2026.
PedidosYa is now using Horizon not as a one-time diagnostic, but as a continuous discovery engine, running discoveries across Tax, Collections, and PQP, with a backlog of prioritized improvements ready to execute each quarter.
Key Results
+31
Insights Generated
14+
Employees Interviewed
5+
Finance Areas in Discovery
| Process Area | Before | After | Impact |
|---|---|---|---|
| Rider Wallet Adjustments | 3–4 h/week of manual retries, ~2,400 riders affected weekly | Automated retry logic identified and mapped | Weekly rework eliminated at scale across markets |
| EDC–SAP Reconciliation (Bolivia) | 1,000+ discrepancies/week, ~2 h analysis vs. 40 min in other countries | Automated matching solution identified | Parity with other-market performance |
| Payout Report Generation | ~30 h/month manual extraction from BigQuery + Excel consolidation | Automated extraction and report generation roadmap defined | Full month of capacity recovered |
| Partner Billing Control | 4–5 h/week, VLOOKUP across SAP/Query/Looker + BackOffice verification for 25,000 partners | Cross-referencing automation identified | Weekly reconciliation time eliminated |
| Banesco Payment Reports | ~40 h/month due to download failures and manual tip processing | Systematic fix identified for Kinpos download failures | ~40 h/month recovered |
| Discovery Process | Manual interviews, fragmented visibility, no quantified waste baseline, covering <1% of workforce | 14+ async AI interviews, 31 findings across 2 teams in 3 months | Full operational picture delivered without calendar disruption |
| Discovery Coverage | Single pilot scope (2 teams) | Expanded to Tax, Collections, PQP, with Rider Care planned | Scalable, continuous improvement program across Finance |
What People Said
“The interview compilation feature saves significant time. It automates a discovery process that would otherwise take months.”
Guido Foussats, Senior Manager Logistics Performance, PedidosYa
“It was really good. It gave us a clear summary of our processes from a completely different angle.”
PedidosYa Team Member, Rider Payments
“It worked great. The questions are logical and the voice recognition picks up everything accurately.”
PedidosYa Team Member, Partner Payments