For organizations pursuing GenAI, success with applications, system integration, data platforms, consulting, tooling, and managed services often depends on the strength of the partner network. In fact, BCG and Amazon Web Service’s third annual GenAI Partner Survey of more than 1,100 organizations found that 75% view partners as major contributors to ROI, and 85% plan to expand partner engagement.
Getting the most out of these partnerships requires understanding how needs evolve during the GenAI journey and recognizing that each partner type has particular strengths along the way. BCG and AWS have identified five distinct phases, from early exploration of GenAI use cases through agentic AI deployment. Companies should develop a partner network to access the right capabilities at the right time and capture GenAI’s promise of operational efficiency, productivity gains, and revenue growth.
Early stages require partners who can secure rapid pathways to ROI and build organizational momentum; mid-stages demand execution-focused capabilities to scale implementations and optimize costs; advanced stages need specialized expertise in orchestration, governance, and autonomous system design to unlock transformational outcomes such as revenue growth and new customer acquisition.
The Five Phases of the GenAI Journey
Most organizations have clear GenAI objectives: 35% expect 3:1 or better ROI within one to two years, and approximately 94% leverage multiple partner types throughout their GenAI journey to attain their ROI goals.
Our research shows how engagement with different partner types evolves across the five-phase maturity curve. For example, the use of classic business consulting partners follows a U-curve: high engagement in the early stages when designing strategy, low engagement during execution, then renewed engagement to address the complexities around agentic transformation. Meanwhile the use of systems integrators peaks at 46% when operational objectives dominate. (See Exhibit 1.)
Stage 1 Experimenting: Identifying the Right Use Cases. During this first stage, organizations identify GenAI opportunities and build the business case for investment. In 2025, 71% of organizations reported engaging partners at this stage, down from 90% in 2024. While the vast majority of organizations continue to seek outside help in this stage, this decline suggests internal expertise is becoming more mature and widespread, allowing companies to conduct exploration more confidently on their own.
Partners can bring a powerful outside-in perspective on how organizations should pursue GenAI: identifying use cases, measuring value with scenario-based modeling, prioritizing the use cases, and aligning teams cross-functionally. Importantly, partners also help identify critical dependencies and adjacent investments that are often overlooked and can block value realization, including data readiness, platform and integration foundations, operating model and governance changes, and workforce enablement.
Stage 2 POC and Piloting: Proving Value and Building Momentum. The proof of concept (POC) stage focuses on validating what’s achievable from productivity gains, operational improvements, or competitive advantages, and then building organizational momentum and securing budget commitments for broader deployment. In 2025, overall partner engagement peaked during the POC phase (87%), followed closely by piloting (86%), a sign that organizations still do not have this specialized expertise in house.
Engagement with consulting and advisory partners specifically climbs to 39% during the POC stage as companies select the tech stack and manage cloud resource costs. Then, as the company moves from POC to piloting, engagement with systems integrators rises to 41%. Organizations need partners who can rapidly stand up POC environments, provide access to models and tools (30% cite this as the top value-adding activity across all partners), and offer expertise in cloud infrastructure setup. As pilots progress, systems integrators shift focus to faster model-training cycles and tailored support services, helping to prove that operational efficiency or productivity gains justify scaling.
Organizations need partners who can rapidly stand up POC environments, provide access to models and tools, and offer expertise in cloud infrastructure setup.
Partners add critical value by ensuring POCs are designed to advance an organization's broader GenAI agenda and they are not treated as isolated technical experiments. As Randall Hunt, CTO at Caylent explains: "POCs are particularly effective in scenarios such as early-stage AI adoption, testing net-new or high-uncertainty use cases, validating data readiness and model feasibility, or pressure-testing cost, latency, and security assumptions before scaling."
This approach was on display in Caylent's work with Symmons Industries, where they infused generative AI into a water-usage monitoring platform to detect leaks and potential plumbing failures. The AI-powered solution saved one Symmons customer 140 million gallons of water (approximately $1.5 million) within seven months. Symmons expects that aggregate savings across its customer base will exceed 5 billion gallons over the next five years.
At the same time, as organizations scale GenAI across multiple use cases, partners help introduce alternative evaluation approaches such as standardized readiness assessments, architectural patterns, and value frameworks. These approaches enable faster decision-making without requiring a standalone POC at every stage.
Stage 3 Deploying: Capturing Operational Gains at Scale. At this stage, use cases must deliver measurable business impact at the enterprise level. Top goals include operational and process optimization (35% of organizations say this is the top priority), followed by improved labor productivity (33%), and then data-driven insights (32%). The typical target for ROI in an organization is between 2:1 and 3:1 returns within one to two years.
Overall partner engagement drops slightly at this stage, although engagement with systems integrators is still strong (36%) as the focus shifts toward day-to-day execution. Managing cloud costs becomes the top priority, followed by faster model-training cycles and tailored support services. Partners help companies address key questions, including: How do we capture productivity gains without exploding our cloud budget? How do we integrate with existing systems to drive operational efficiency? How do we support dozens of teams deploying use cases that target different outcomes?
Partners help companies address key questions, including: How do we capture productivity gains without exploding our cloud budget? How do we integrate with existing systems to drive operational efficiency? How do we support dozens of teams deploying use cases that target different outcomes?
As organizations deploy GenAI at scale, data platform partners reach peak importance. This helps ensure access to high-quality, scalable data that can support enterprisewide models and dozens of concurrent use cases. At this stage, unified access to data, analytics, and AI becomes essential. GenAI outcomes depend on the ability to consistently prepare, govern, and operationalize data across teams without introducing fragmentation or excessive cost.
For example, Databricks recently worked with Condé Nast, a global media company, to migrate more than 800 media properties to a unified AWS cloud infrastructure, then built a centralized data-analytics platform to process data from all global brands. This transformation allowed Condé Nast to shift from opinion-driven to data-driven strategies backed by unified audience insights, achieving $6 million in cost savings while gaining the agility to deliver personalized content experiences to a diverse audience.
Stage 4 Fully Operationalizing: Sustaining and Optimizing Performance. Once GenAI systems are in production, the focus shifts to sustaining efficiency gains and expanding impact. In 2025, overall partner engagement at this stage was 85%, while the partner mix shifts toward those with capabilities focused on operational optimization and sustained delivery for steady-state operations. In this phase, AI tooling partners (a rapidly growing category) play an increasingly important role (33%) alongside systems integrators (30%) and managed service providers (27%).
For example, partners like Dynatrace provide production-grade observability for generative AI architectures, enabling teams to monitor model performance and cost in real time as usage scales. Other partners, like Arize AI, help ensure that the data feeding GenAI systems remains accurate, stable, and fit for purpose over time, surfacing issues early before they degrade outputs or require costly rework. Elsewhere, partners such as Elastic support GenAI applications by making enterprise knowledge easily searchable and retrievable, helping models generate responses grounded in trusted data rather than generic or outdated information.
This is the stage where companies report the most need for partner improvement. Only 58% are satisfied with their partner mix, with the top pain points being data management and preparation (21%) and insufficient ongoing support (20%). Given the rate of technological change, partner offerings and capabilities must evolve to remain in scope and help organizations sustain and expand impact.
Stage 5 Agentic Transformation: Unlocking More Extensive and Impactful Outcomes. In the final stage, autonomous agents promise to move organizations beyond basic productivity improvements to true business model transformation. We found that 34% of organizations anticipate that GenAI will deliver value by accelerating R&D and faster go-to-market, 34% expect GenAI will deliver value by improving the customer experience, and 33% believe value will come from growing customer lifetime value. Realizing this value requires a mix of speed and strategy; it’s not only doing work faster, it's also doing entirely new work that wasn't possible before.
In the final stage, autonomous agents promise to move organizations beyond basic productivity improvements to true business model transformation.
In the midst of this disruptive moment, we find that the majority of enterprises become more reliant on partners for agentic transformation than for more basic GenAI implementation. To deliver here, partners need to support companies across orchestration, governance, and translating operational KPIs into bottom-line financial impact.
A recent example is Loka's work with Experity, a health care technology company serving urgent care and occupational health providers. Experity partnered with Loka to build Care Agent, an agentic AI solution that goes beyond basic voice-to-text transcription. "We partnered with Loka to design and vet an architecture that turns clinical conversations into discrete data that’s usable downstream, including for billing," says Chris Hoffman, chief architect at Experity.
Sushmitha Regulapati, VP of health care and life sciences at Loka, adds: "Health care teams are overwhelmed by administrative work that pulls them away from patients. Agentic AI can meaningfully reduce that burden.”
"Health care teams are overwhelmed by administrative work that pulls them away from patients. Agentic AI can meaningfully reduce that burden.”
The solution leveraged AWS infrastructure to scale rapidly and reach the market in an aggressive three-to-four-month time frame. AWS HealthScribe is now live with approximately 2,000 providers, reducing visit time by 5–15 minutes per encounter and reclaiming up to an hour and a half of personal time daily for some providers. In other words, GenAI can improve operational efficiency and address critical workforce challenges. "It's having a meaningful impact on provider burnout," Hoffman notes.
Shifts in Purchasing Criteria and Discovery
In addition to identifying specific capabilities needed during each phase of the GenAI journey, our survey also identified changes in the key purchasing criteria used for partner selection. More organizations now recognize that they need partners with deep technical expertise to capture GenAI’s potential. (See Exhibit 2.) “Experience with the tech stack” leads in driving partner selection, surpassing the prior year’s top criterion: “Quality customer service and support.”
There is also a shift in partner discovery methods: word of mouth matters. To find service partners who focus on operational efficiency and process optimization, organizations rely on peer recommendations first. When looking for technology partners (focused on technical implementation and infrastructure), organizations rate cloud service provider recommendations highly, given their perspective on ecosystem integration and capabilities necessary to capture technical outcomes.
Maximizing Value from AI Engagement
With all these shifts year over year, we have identified several actions organizations can take to improve their GenAI journey.
Define your objective, then match partners. If you're pursuing employee productivity gains, you need partners who excel at change management and adoption. If you're targeting operational efficiency, you need partners who understand process complexities. If you're after revenue growth, you need partners who understand your customers’ journeys. If you don’t know, you need a partner who can help you set this strategy.
Build a roster aimed at capturing future value. If you're capturing productivity gains today, start conversations with partners who can help achieve operational transformation tomorrow. If you're optimizing operations today, connect with partners who can help you pursue revenue growth through agentic capabilities. The organizations achieving the strongest returns maintain relationships across the spectrum even when one objective dominates their current focus.
Address the top pain points blocking impact. Data management (21%) and ongoing support (20%) are the leading pain points. These directly impact value realization. You cannot achieve data-driven insights without strong data management, and you cannot sustain productivity gains without robust ongoing support. If your current partners struggle in these areas, ambitions will be frustrated. Most organizations require multiple partners to tackle the range of challenges.
Evaluate partners against business impact, not just technical delivery. Selection criteria should focus on proven results. Prioritize accreditation and competency certification, clear proof of ROI and business impact, and technical expertise validated through customer references and case studies. A partner with impressive technical credentials but no evidence of delivering your specific objectives (operational efficiency, revenue growth, etc.) may not be the right fit.
A partner with impressive technical credentials but no evidence of delivering your specific objectives (operational efficiency, revenue growth, etc.) may not be the right fit.
Track value realization, not just deployment metrics. Measure partner contributions against your actual objectives: Did they help achieve the 3:1 ROI you targeted? Did productivity improve by the expected margin? Are you capturing the operational efficiencies in your business case? Partner performance should be evaluated on measurable financial impact, not tasks completed. Design contracts with milestone-based payments staggered over time.
The organizations achieving the strongest GenAI returns share a common approach: they've aligned their partner strategy with their technology roadmap. They recognize that the objectives defining each maturity stage differ substantially, and each requires specialized expertise. This, in turn, creates patterns across the GenAI journey. With this in mind, companies need to evolve their partner network intentionally and deploy partners specifically.
