How to Manage a Hybrid Squad of AI Agents and Colombian SDRs

· Manuel · 13 min read · Innovation

For decades, US companies outsourced their low-tier customer service and cold calling to offshore BPOs in India or the Philippines to save on labor costs. In 2024, the Nearshore model (specifically Colombia) gained massive market share because of time-zone alignment and cultural affinity.

But by 2026, the entire concept of "I need 50 cheap bodies in a room making 400 dials a day" is officially obsolete. If a task requires rote repetition and a strict script, a vAPI Voice Agent will outwork a human 100 to 1 at a fraction of the cost.

However, as we've established in the Human-in-the-Loop Doctrine, you cannot eliminate humans from the closing sequence. At GSD 500, we pioneer the Hybrid Squad Model. We do not hire 50 offshore dialers; we hire 5 elite, hyper-compensated Colombian AEs (Account Executives), and we give them an army of AI assistants.

Here is how we manage this hybrid architecture on an operational level.

1. The Squad Micro-Structure

Instead of a hierarchy of "Manager -> Team Lead -> SDR," we organize into Pods. A single Pod consists of:
  • 1 Elite Colombian Human AE (The Closer, sitting in Bogota or Medellín).
  • 10 AI Voice Agents (The Grinding Frontline).
  • 1 n8n Workflow State Machine (The Orchestrator).
  • 1 QA Shadow Bot (The Auditor).

    • The Human AE acts more like an Air Traffic Controller than a traditional salesperson.

    2. The Command Center (Supabase & React)

    The Human AE does not dial phones natively. They sit inside the custom GSD 500 React Dashboard.

    On their left monitor, they have a live scrolling feed (powered by Supabase Realtime) of the 10 AI Voice Agents making hundreds of simultaneous cold calls across the US timezone.

    When an AI detects extreme buying intent from a prospect (e.g., the prospect says, "Actually I am looking for a roofer today"), the row flashes green on the AE's screen. The AI executes a warm SIP transfer, and the call instantly drops into the Human's headset. The AE is seamlessly fed the live translated transcript so they know exactly what was said 10 seconds prior.

    3. Asymmetric Compensation

    Traditional offshore BPOs pay notoriously low flat salaries. They rely on high turnover and low expectations. In the Hybrid Squad model, our human reps are handling massive volume because they only talk to high-intent transfers. Their close rate is astronomically higher than a standard SDR.

    We compensate them fiercely with US-style performance bonuses. This creates extreme loyalty and retention. When your Colombian AE is closing 4x the volume of a traditional BPO because their AI assistants are filtering out the garbage, everybody wins.

    4. Continuous Feedback Loops (The Daily Standup)

    AI cannot fix a bad script on its own. Every morning at 8:30 AM EST, the Human AE reviews the "QA Shadow Bot" report. This report flags the top 5 calls from yesterday where the AI failed to overcome an objection.

    The Human AE acts as the Prompt Engineer for their specific pod. They analyze the failure. Human AE: "Ah, the AI didn't know how to handle the objection about the new Texas EPA regulations." The AE updates the Supabase Vector knowledge base with a new 2-paragraph memo explaining the Texas EPA shift. In 30 seconds, the 10 AI Agents instantly update their behavior across thousands of new calls.

    5. Managing Cultural Nuance

    AI is phenomenal at logic. Humans are phenomenal at reading the room. If an AI agent is transferring a prospect who sounds highly stressed or aggressive over the phone (detected via sentiment analysis), the UI alerts the Colombian AE: WARNING: Client Sentiment is HIGH TENSION.

    The AE knows to answer the transfer with a slow, calming voice, bypassing normal small talk and moving straight into solution-oriented empathy.

    This is the future of the agency BPO model. An orchestra of bleeding-edge silicon doing the grunt work, led by highly intelligent, well-compensated human conductors dealing exclusively in empathy and trust. `,

    // ========================================== // SPANISH CONTENT // ========================================== contentEs: ` El modelo tradicional de BPO donde "necesito poner a 50 personas baratas a hacer llamadas en frío todo el día" está muerto. Una IA hace ese trabajo repetitivo por una fracción del costo y a mayor velocidad.

    Pero como establece nuestra Doctrina, el humano es irreemplazable al final del embudo, en el cierre. En GSD 500 fuimos pioneros del Modelo de Escuadrón Híbrido (Hybrid Squad). No contratamos 50 agentes de nivel bajo; contratamos 5 Ejecutivos de Cierre de Élite en Colombia, y les asignamos un ejército digital de Bots. Así se gestiona:

    1. La Micro-estructura del Escuadrón

    Organizamos a las personas en Células (Pods). Una célula es:
  • 1 Ejecutivo Humano (Ubicado en Bogotá o Medellín).
  • 10 Agentes de Voz de IA peleando en el frente.
  • 1 Máquina de Estado automatizada (n8n).
  • 1 Bot de Auditoría o "Sombra".

    • El humano funciona más como un controlador de tráfico aéreo que como un marcador de teléfono tradicional.

    2. El Centro de Mando (Supabase + React)

    El ejecutivo no marca teclas. Mira nuestro panel central interactivo. En la pantalla izquierda, ve el estado en tiempo real de los 10 Bots disparando cientas de llamadas en vivo por todo EE. UU.

    Si el Bot capta alta intención de compra, la fila parpadea en verde. La IA ejecuta una transferencia inteligente y el sonido cae directamente a los auriculares del humano, junto a una transcripción del texto de lo que la IA y el cliente acaban de hablar.

    3. Compensación Asimétrica de Élite

    Los BPOs tradicionales de atención al cliente pagan el mínimo para aguantar la alta rotación. Nosotros no sufrimos esto. Los humanos del escuadrón son la estrella y lidian solo con clientes filtrados y listos. Cierran a un ritmo abrumador y son compensados ferozmente con bonos por agresivo rendimiento (esquema estadounidense), lo que genera extrema lealtad.

    4. Retroalimentación Matutina Inmediata

    La IA no se repara sola si algo del mercado cambia. Cada mañana, el humano revisa el informe del "Bot Sombra" con las 5 peores llamadas de la IA donde no logró superar la objeción del prospecto. Al igual que un gerente corrige a un empleado nuevo, el Ejecutivo Humano reescribe el documento del prompt o la base de conocimientos de los bots, y la IA queda actualizada instatáneamente.

    Una sinfonía de silicio masivo haciendo el trabajo sucio, liderada por humanos altamente compensados, enfocados estrictamente en la empatía y la confianza.