Managing Human + Agent Teams
Ross Sylvester, Co-Founder & CEO, Adrata | Feb 2026 | ~11 min read
The CRO's job just forked. You still manage people. Now you also manage machines. The playbook for doing both well doesn't exist yet — so here's the one we're writing in real time.
The Fork in the Road
Sometime in the last eighteen months, every revenue leader I know hit the same inflection point. They looked at their SDR team, looked at their pipeline targets, looked at their budget, and realized the math had fundamentally changed. Not incrementally. Structurally.
The old equation was simple: more pipeline requires more headcount. Need 40% more qualified meetings next quarter? Hire six more SDRs. The new equation is different: more pipeline requires better orchestration of humans and agents working together on the same funnel.
This is not a theoretical shift. Gartner predicted that 75% of B2B sales organizations would augment traditional sales playbooks with AI-guided selling solutions by 2025.1 That prediction landed roughly on time. Outreach's 2025 Sales Data Report documented that organizations using AI in their sales pipelines are seeing a 20% increase in pipeline volume and a 30% improvement in lead conversion rates.2 And the economics are decisive — AI SDR platforms are delivering 4-7x higher lead-to-meeting conversion rates while reducing customer acquisition costs by up to 70%.3
But the technology is the easy part. The hard part is the management problem nobody prepared you for: how do you actually run a team where some of your "reps" are humans and some are software?
The Productivity Gap Is Real — and Growing
Let's start with the numbers, because the numbers are what force the conversation.
A traditional sales rep spends 25-28% of their time on actual revenue-generating activities.4 The rest vanishes into CRM updates, internal meetings, prospecting research, email composition, and administrative work. Bain's 2025 Technology Report found that AI could effectively double active selling time by automating routine tasks — suggesting the current model wastes roughly half of every rep's productive capacity on work that delivers zero direct revenue value.4
AI agents don't have this problem. They don't context-switch. They don't attend all-hands meetings. They don't spend 45 minutes updating Salesforce after a call. They prospect, enrich, sequence, and qualify around the clock.
The cost differential is equally stark. The annual fully-loaded cost of a human SDR runs $110,000-$150,000. An AI SDR agent costs $28,000-$35,000 per year — a 70-75% reduction — while operating 24/7 with zero ramp time.5 When you factor in the 72% of human time spent on non-selling work, the cost per effective selling hour for a human SDR is roughly $240-320. For an AI agent, that number approaches single digits.
Sales reps using AI tools are 3.7x more likely to hit quota.6 Not marginally more likely. Nearly four times more likely.
These numbers don't argue for eliminating humans. They argue for rethinking what humans should be doing.
The New Division of Labor
The highest-performing revenue teams in 2026 are converging on a simple framework: agents handle repetition, humans handle relationships.
This isn't a slogan. It's an operational architecture. Here's how it breaks down in practice:
What agents do better:
- Initial outreach sequencing at scale (hundreds of personalized touches per day)
- Lead enrichment and data hygiene (Clay's AI research agents more than doubled enrichment coverage from the low 40% range to the high 80% range at OpenAI)7
- Intent signal monitoring across dozens of data sources simultaneously
- Meeting scheduling and calendar coordination
- CRM data entry and pipeline hygiene
- Follow-up sequencing after no-shows or stalled conversations
What humans do better:
- Reading emotional subtext in live conversations
- Navigating internal politics within a buying committee
- Building genuine trust with economic buyers
- Handling novel objections that require creative problem-solving
- Negotiating complex commercial terms
- Making judgment calls when the playbook doesn't apply
The mistake most CROs make is treating this as a binary — either automate the function or keep it human. The right model is a relay race. The agent runs the first leg: identifying prospects, enriching data, personalizing outreach, qualifying interest. The moment a prospect signals genuine engagement — a reply with substance, a question that reveals a real problem — the baton passes to a human who can do what no agent can: build a relationship.
Monday.com's 2026 analysis of hybrid sales teams found this pattern consistently: AI handles sending initial outreach sequences, tracking engagement signals, and scheduling discovery calls, while human SDRs take over once prospects show genuine interest, handling objections, conducting needs assessments, and building authentic connections.8
The handoff point is everything. Too early, and you're wasting human time on unqualified leads. Too late, and the prospect feels like they've been talking to a machine (because they have). The best hybrid teams are obsessive about tuning this transition.
Redesigning Quotas for Hybrid Teams
Here's where most CROs get stuck: how do you set quotas when an agent is generating the top of your funnel?
The old SDR model was straightforward. An SDR was responsible for X qualified meetings per month. An AE was responsible for Y pipeline and Z closed revenue. The quotas were set against headcount, and the math was predictable within a band.
In a hybrid model, the SDR function is partially or fully automated. This creates three quota design problems:
Problem 1: Activity metrics become meaningless. If an AI agent is making 500 personalized outreach attempts per day, measuring "calls made" or "emails sent" tells you nothing about human performance. Shift incentives toward outcomes — pipeline generated, meetings held, revenue closed — and AI becomes a tool for hitting numbers rather than a threat to the activity scorecard.
Problem 2: Individual attribution gets murky. When an agent enriches a lead, personalizes the outreach, books the meeting, and then hands it to an AE who closes the deal, who "sourced" the pipeline? This matters enormously for compensation. The answer most teams are landing on: stop attributing to individuals and start attributing to the system. The AE's quota is set against the pipeline the system delivers. The AE's job is conversion, not sourcing.
Problem 3: Quotas need to rise — but thoughtfully. If your agents are generating 3x the qualified meetings your SDR team used to produce, your AEs can handle more pipeline. But "more" doesn't mean "3x more." Human bandwidth for relationship-based selling doesn't scale linearly. The best hybrid teams are raising AE quotas by 30-50% — not 200% — and using the surplus pipeline to improve selectivity. Better leads, not just more leads.
One manufacturing company that shifted to a rolling six-month quota system based on AI analysis of their pipeline data saw 23% lower rep turnover.6 Quotas that reflect reality rather than aspiration retain talent. And in a world where your best humans are more valuable than ever — because they're the ones the agents can't replace — retention is a strategic priority.
Compensation Design in the Hybrid Era
Compensation is where theory meets human psychology, and where most hybrid team rollouts fail.
The fundamental tension: if reps are compensated for activity (calls made, demos delivered), AI feels threatening because it handles those activities. If reps are compensated for outcomes (pipeline generated, revenue closed), AI becomes an accelerant.
Here's the compensation framework emerging from the companies doing this well:
For AEs in hybrid teams:
- Base salary remains 40-50% of OTE (unchanged)
- Commission shifts entirely to closed revenue and expansion (no credit for sourcing)
- Add a multiplier for deals involving complex, multi-stakeholder selling (rewarding the work agents can't do)
- Introduce a "system contribution" bonus for reps who actively improve agent workflows — training data, feedback loops, playbook refinements
For the remaining human SDRs (now fewer, more senior):
- Retitle the role: these aren't SDRs anymore. They're "Pipeline Strategists" or "Demand Architects"
- Compensate on qualified pipeline value, not meeting volume
- Add compensation for agent oversight quality — the humans who QA agent output, catch errors, and improve targeting
- Higher base, lower variable split (70/30 instead of 60/40), reflecting that their value is judgment, not volume
For agent managers (a new role):
- This is the role Harvard Business Review described in February 2026: a dedicated position responsible for supervising AI agents as if they were employees.9 At Salesforce, Zach Stauber manages a fleet of generative AI agents across support, sales, and marketing on their Agentforce platform. His job is monitoring how agents are working, learning, and adapting — the same way a traditional manager might walk the floor and check in with a struggling employee.
- Compensate agent managers on system-level metrics: pipeline throughput, conversion rates, cost per qualified meeting, agent accuracy scores
- This role sits between RevOps and Sales Management. It requires both technical fluency and commercial judgment.
The critical principle: nobody should feel punished for the existence of AI agents. If your comp plan makes humans feel like they're competing with machines, you've already lost your best people.
What OpenAI Learned Running Hybrid Teams
OpenAI's own GTM organization is one of the most instructive case studies in hybrid team management — because they're simultaneously selling AI and using AI to sell.
When OpenAI's go-to-market team tripled in size while launching new products nearly every week, they hit a structural problem: sales reps were spending an hour preparing for thirty-minute calls, bouncing across dozens of systems to piece together context. Customers sent hundreds of product questions weekly that bogged down subject-matter experts and slowed deals.10
Their solution was the GTM Assistant — an internal AI agent that handles meeting prep, product Q&A, and account research. One OpenAI rep exchanges 22 messages per week with the GTM Assistant across daily briefs, recaps, and Q&A, and reports a 20% productivity lift — roughly one extra day per week to spend with customers and manage a larger book of business.10
But the real insight was how they trained it. Top-performing reps worked directly with the GTM Assistant, shaping what "great" looks like in meeting briefs and product responses. Their expertise trained the system so that every evaluation, correction, and improvement scaled the habits of the best sellers across the entire organization.10 The AI didn't replace the best reps. It cloned their preparation habits and gave them to everyone.
For lead enrichment, OpenAI used Clay's AI research agent to automate and scale custom GTM research — mimicking the process of their best sales representatives. They started by automating what top sellers already did: visiting company websites, checking LinkedIn pages, and pulling key information like recent developments and revenue figures. The result: enrichment coverage more than doubled, from the low 40% range to the high 80% range, which meant leads were properly scored, routed, and responded to. Individual sellers ran up to 150 lead enrichments on busy days, particularly during quarter starts when building pipeline.7
The lesson for CROs: don't start with the technology. Start with your best reps. Document what they do differently. Then build agents that replicate those patterns at scale while freeing those same top performers to focus exclusively on the high-judgment work that moves deals.
The Five Leadership Skills CROs Need Now
The CRO role is splitting in two. Half of it is still the people-leadership job you were hired for — coaching reps, running forecasts, managing up to the board. The other half is a systems-design job you probably never trained for. Here are the five skills that matter most:
1. Workflow Architecture. You need to think in systems, not headcount. Which activities in your revenue process are repetitive and rule-based? Those go to agents. Which require judgment, empathy, and creativity? Those stay with humans. The CRO who can design these workflows — and redesign them monthly as the technology improves — will outperform the CRO who's still thinking in terms of hiring plans.
2. Agent Performance Management. HBR's February 2026 piece made the case that companies need dedicated "agent managers," but in most organizations the CRO sets the standards.9 You need to define performance metrics for your agents the same way you define them for your reps: conversion rates, response quality, handoff accuracy, pipeline throughput. And you need to audit agent outputs for accuracy and brand alignment — because an agent that sends 500 bad emails in a day does more damage than an SDR who sends 50.
3. Change Leadership. Your existing team is scared. They've read the same headlines about AI replacing salespeople. Your job is to articulate a clear narrative: we're not replacing you, we're promoting you. The repetitive work you hate is going to machines. The complex, relationship-driven work you're great at is now your entire job. And your quota reflects that shift.
4. Data Fluency. You don't need to be a data scientist, but you need to understand what data your agents are consuming, where it's accurate, and where it fails. Boards and CEOs are shifting from experience-based evaluation ("Have you done this before?") to skills-based evaluation ("Can you do what the moment requires?").11 Data fluency is now table stakes for the role.
5. Experimentation Velocity. The organizations that treat AI as "set it and forget it" will be outpaced by competitors who iterate weekly on their AI systems.12 The best CROs are running continuous experiments — testing new agent workflows, measuring results, adjusting prompts, refining handoff criteria. This is not a one-time transformation. It's a permanent operating rhythm.
The Uncomfortable Truth
Here's what I tell every CRO who asks me about this transition: the hybrid team isn't coming. It's here. By the end of 2026, 83% of executives anticipate AI agents will autonomously execute actions based on operational metrics and transaction histories.3 If you're still running an all-human SDR team, you're not being thoughtful. You're being slow.
But the opposite extreme is equally wrong. Gartner's own research predicts that by 2030, 75% of B2B buyers will prefer sales experiences that prioritize human interaction over AI.13 The buyers want to talk to a person — just not for the parts that don't require a person.
The CROs who win this transition will be the ones who understand that managing human + agent teams isn't about choosing between humans and machines. It's about designing a system where each does what it does best, where the handoffs are seamless, where the compensation reflects the new reality, and where the humans on your team feel like they've been elevated, not replaced.
That's not a technology problem. That's a leadership problem. And it's the most important one you'll solve this year.
References
Footnotes
-
Gartner, "75% of B2B Sales Organizations Will Augment Traditional Sales Playbooks with AI-Guided Selling Solutions," 2023. ↩
-
Outreach, "Sales 2025 Data Report: Trends, AI & Sales Benchmarks," 2025. ↩
-
MarketsandMarkets, "How Agentic AI in Sales is Redefining SDR Productivity," 2025. ↩ ↩2
-
Bain & Company, "2025 Technology Report: AI and Sales Productivity," 2025. ↩ ↩2
-
Monday.com, "Will AI Replace SDRs? The Data on Hybrid Sales Teams in 2026," 2026. ↩
-
SalesGlobe, "Designing Sales Compensation Plans in a Data-Driven Environment," 2025. ↩ ↩2
-
Clay & OpenAI, "How OpenAI Is Scaling Their GTM Motion with Clay," 2025. ↩ ↩2
-
Monday.com, "AI SDR Agents: The Complete Guide for Sales Teams in 2026," 2026. ↩
-
Harvard Business Review, "To Thrive in the AI Era, Companies Need Agent Managers," February 2026. ↩ ↩2
-
OpenAI, "Driving Sales Productivity and Customer Success at OpenAI," 2025. ↩ ↩2 ↩3
-
TalentFoot, "The Top AI Leadership Skills in 2026: Salary Premiums and Case Studies," 2026. ↩
-
SaaStr, "AI, Sales + GTM in 2025/2026: This Changes Everything," with Jason Lemkin and Kyle Norton, 2025. ↩
-
Gartner, "By 2030, 75% of B2B Buyers Will Prefer Sales Experiences That Prioritize Human Interaction Over AI," August 2025. ↩