Matt Wolodarsky, a principal program manager at Microsoft, has led an early access program of more than 90 customers, testing out how no-code agents in SharePoint can drive value for their businesses over the past 18 months.
Recent headlines suggest that organizations are wasting money on AI with little to show for it. What we’ve observed is more nuanced, and far more actionable.
In the SharePoint AI Early Access Program, encompassing 90+ early adopter organizations piloting and rolling out agents built in SharePoint broadly, one pattern stands out: the technology is powerful, but adoption only scales when people discover where agents fit into their work, feel supported as they experiment, and build on early wins so adoption deepens into new habits that change how work gets done.
Agents that users build in SharePoint act like subject matter experts grounded on your organization’s content, that can be put to work across your workflows and business problems.
For example, some customers use agents to help salespeople draft tailored proposals informed by prior winning proposals, help project managers learn from past projects, and review marketing deliverables to ensure they are compliant with local market advertising laws.
From the experiences of our private preview customers, five adoption practices consistently make the difference between early excitement that fizzles out and adoption that delivers lasting business impact. To make these adoption best practices easier to act on, I’ll also introduce two new meta-agents that you can create in your tenant right away to help your organization put these practices into action.
1. Inspire with scenarios so people know where to start
At this early stage of the enterprise adoption curve, the concept of delegating work to AI agents is unfamiliar for most employees. It’s a shift in mindset, and many people don’t yet understand what role agents should play in their workflow. As a result, adoption stalls and the value of AI investments to the organization does not get fully realized.
That’s why organizations getting real value from AI go beyond telling people that agents are useful – they inspire employees with practical examples tied to their role, goals, and challenges.
One example comes from Amey, a leading provider of engineering, operations, and decarbonization solutions. They’re using SharePoint agents to help employees quickly find accurate health and safety information, ensuring their teams stay safe while working on complex projects. By anchoring their agent solutions in a tangible challenge like occupational health and safety, Amey sparked exploration that led to the creation of a critical SharePoint agent to help on-site employees stay safe on the job.
To help you inspire your own teams with role-relevant examples, we’ve published the Agent Scenario Playbook - a collection of role-based examples you can adapt.
You can also create a meta-agent called the Agent Ideation Partner in your tenant to help users discover high-value scenarios aligned to their roles, goals, and context.
Inspiring users with relevant scenarios is the first step. Once those first use cases prove valuable, the next unlock is expanding beyond simple “question and answer” (Q&A) scenarios to broader workflows.
2. Expanding beyond Q&A
Organizations that go beyond Q&A type scenarios, realize even greater value from agents. I’ve helped many users reimagine how agents can apply creativity and reasoning across multiple steps of a workflow – not just answer questions.
For example, one pharmaceutical company began with a simple Q&A agent that Clinical Writers used to ask guidance-related questions during trial planning. They soon realized that more of the clinical trial design workflow could be streamlined by creating an agent through SharePoint that automatically generates a first-draft trial timeline – grounded on current regulatory guidance, company standards, and key calendar considerations such as holidays and major events. Writers then review and improve the draft, ensuring compliance through human oversight.
People don’t just want answers – they want outcomes. Across industries, customers are using agents grounded on gold standard content and past examples to generate tailored content. One standout case is the RFP (“Request for Proposal”) response agent featured in the Agent Scenario Playbook, which helps proposal writers craft brand-aligned, high-quality responses in a fraction of the time.
Creating an "RFP response agent" from SharePoint helps proposal writers craft quality responses in a fraction of the time.
Some customers are taking this concept one step further, building agents grounded in proposal evaluation criteria to review the draft proposal and provide feedback to help make the proposal more likely to win the bid. From agents that review marketing copy for compliance, or agents that perform greenwashing reviews, agents – grounded in tailored SharePoint content – are proving to be powerful assistants across a wide range of content creation workflows.
3. Help everyone write clear, detailed instructions to get better results from agents
As organizations integrate agents into daily workflows, success increasingly depends on how clearly users can express what they want the agent to do, and how it should be done to meet the task requirements.
Effective instruction design requires the right balance of detail, context, and precision so the agent can deliver consistent, high-quality responses. In some cases, success also depends on knowing when to divide a broad or complex scenario into multiple specialized agents, rather than relying on a single, all-purpose agent to manage every step of a workflow. Much like mastering search or spreadsheet formulas in previous eras, agent instruction authoring is becoming an essential competency that empowers employees to get reliable results from AI agents.
Employees need practical support to develop this new skill. You can deploy the Instruction Wizard Agent in your tenant – a meta-agent designed to help users craft clear, high-quality instructions for their own SharePoint agents.
Organizations that invest in helping their people develop and apply instruction design skills are seeing faster adoption, better agent responses, and greater confidence in using agents productively.
4. Refine and iterate instructions for continuous improvement
The best-performing agents I’ve seen weren’t “one and done” builds. They became effective through a cycle of testing, real-world deployment, feedback, and iteration.
Agent creators who treat their agents as a living system – refining instructions based on real user behavior and feedback, updating knowledge sources as gaps emerge, and adjusting as business needs evolve – consistently unlock more value than those who stop at the first working draft. To help, creators can use the Knowledge Agent to keep their SharePoint content fresh and relevant, optimizing it for agents to reason over and provide better responses.
Think of launching an agent less like completing a project and more like onboarding a new team member. A new hire needs a job description, training, and ongoing feedback to grow into the role. Agents are no different. By investing in continual improvement – adjusting their “job description”, refining the information and sources they rely on as new lessons are learned or policies change, testing against reality, and incorporating user feedback loops – you can steadily optimize performance, strengthen trust, and drive adoption.
Instructions aren’t static. They’re the single most important lever for turning an agent from a demo curiosity into an adaptable, bar-raising productivity driver. As agents mature through this cycle of refinement, each improvement sharpens performance - reducing errors, broadening coverage, and earning user trust. In time, that steadier performance makes it easier to attribute real efficiency gains to the agent itself – and to show its business impact in clear, defensible ways.
5. Measure ROI to build trust and momentum
In today’s climate of widespread uncertainty about how AI investments translate into real value, measurement is what separates momentum from skepticism. Executive sponsorship doesn’t last on anecdotes. Leaders expect proof of impact before they scale adoption. Establishing clear metrics such as productivity gains, higher quality outcomes, or accelerated time to market turns early promise into credible evidence of business value.
Some of the most credible ROI analysis comes from simple, well-designed measurements grounded in real workflows.
A good example is from how Microsoft’s legal team is using SharePoint agents to help 3,000 marketers quickly find legal and compliance answers. By comparing manual search times to agent-assisted responses, they found marketers worked 2.97× faster — a concrete, defensible proof of business impact.
When teams measure impact through real work and clear results, they don’t just prove value – they build belief. And that belief is what keeps AI momentum alive.
Putting It All Together: The No-Code Agent Adoption Toolkit
Together, these five success factors form the foundation for successful no-code agent adoption. But understanding what drives adoption is only the starting point – real transformation happens when teams put it into practice. To help you get started, we’ve published blueprints for two meta-agents you can create and deploy today.
Agent Ideation Partner
Helps users identify high-impact agent scenarios with clear success metrics. Learn how to build this agent: https://aka.ms/AgentIdeationPartner
Instruction Wizard
Helps users craft and refine effective agent instructions. Learn how to build this agent: https://aka.ms/InstructionWizard
By making these agents available in your environment, you give teams a head start in applying what works and building momentum for adoption at scale.
Ready to go further? Be among the first to explore new SharePoint AI innovations like no-code, custom agents. Express your interest in early access to Private Preview features and help shape the future of intelligent content experiences with SharePoint by nominating your organization for our SharePoint AI Early Access Program at https://aka.ms/JoinSharePointAI
As organizations apply these practices, we’re seeing a new kind of knowledge ecosystem take shape – where SharePoint content and agents work together to help people learn, build on what they know, and get work done faster.
Microsoft