AI in Impact Investing and Development Finance: Practitioner Insights
July 2026
Author: Gigi Alsaadi is a member of the CAFIID Thought Leadership Committee
“The question is no longer whether organizations are using AI, but how they are using it responsibly.”
On June 16, 2026, the Canada Forum for Impact Investment and Development (CAFIID) hosted an online practitioner conversation on AI in Impact Investing and Development Finance. The session brought together professionals working across impact investing, development finance, ecosystem building and other related fields to explore a fast-moving question: how is AI actually being used in the field today?
This practical conversation featured a moderated discussion with Michelle Baldwin, Co-Executive Director of Impact United Academy, and Tony Guo, Portfolio & Investment Selection Principal at Unreasonable Group, and facilitated breakout rooms. These smaller discussions created space for practitioners to share where they are experimenting with AI, what is creating value, and where caution is still needed.
Poll responses from 16 participants provided an additional snapshot of AI adoption, applications, opportunities, and concerns. While the findings reflect this particular group rather than the broader sector, they closely reinforced the themes raised throughout the session.
We share the session’s insights below as aggregated themes, without attribution to specific participants or organizations.
AI is already embedded in day-to-day work
One of the clearest takeaways from the webinar was that AI adoption is no longer theoretical: the key divide is no longer between adoption and non-adoption, but between informal experimentation and more systematic integration. All participants described using AI tools in at least limited ways, covering a wide range of practical activities, including research, due diligence, portfolio monitoring, communications, financial analysis, memo preparation, website development, and impact measurement.
Guo shared how Unreasonable Group is using AI across the investment workflow, not as one general tool replacing people, but as a set of tailored workflows designed around specific tasks. For example, AI can help filter large volumes of potential companies, summarize technical materials, support due diligence preparation, review financial data, assist with investment memos, monitor portfolio news, and draft email responses.
A key point from Guo’s remarks was that AI is most useful when it is designed around a specific workflow. In sourcing, for example, AI can help screen opportunities from databases and newsletters so that teams spend less time manually reviewing lists and more time building relationships with companies. In due diligence, it can help investors understand unfamiliar technical fields more quickly. In financial review, it can help summarize complex models, while still requiring human review to understand the nuance.
Responses from participants reflected this focus on information-intensive work as research and due diligence was the leading area in which organizations were using or interested in using AI (12 mentions), followed by communications and knowledge management (11) and impact measurement and reporting (7).
Some organizations are still using AI informally and individually, while others are beginning to integrate it more systematically. Several participants described using AI as an “analyst” or “thought partner” for early-stage research, benchmarking, identifying red-flags, industry scans, and sensemaking across large volumes of information.
The immediate value is speed, synthesis, and knowledge management
The most immediate opportunities identified by participants were practical: reducing due diligence time and cost, improving impact measurement, democratizing access to investments, and identifying patterns across portfolios.
AI was seen as especially useful for making sense of fragmented information. In many investment and development finance workflows, relevant data sits across many different sources. AI tools can help consolidate those inputs, identify patterns, and produce first drafts or summaries that would otherwise take hours to prepare.
Participants also noted the value of AI-enabled transcription and note-taking. In relationship-driven work, important information often emerges through conversations rather than formal documents. AI note-taking tools can help capture more detail from calls, support better follow-up, and allow teams to focus more fully on the conversation itself.
Context makes AI more powerful, but also more complicated
For many organizations, the next challenge is not simply “how do we use AI?” but “how do we manage AI responsibly at the organizational level?”
AI becomes more useful when it can draw on organizational knowledge and context. Guo noted that centralized information can make AI outputs much stronger because the tool can draw on internal/organizational knowledge. However, this also creates governance questions. The more context an AI tool can access, the more powerful it becomes, but also the more important it is to understand how data is governed and whether it aligns with cybersecurity and confidentiality policies.
Organizations need a much clearer understanding of what information is being shared and what safeguards are in place, while tackling other concerns, like data hosting, anonymization, access, and information security.
Human judgment remains central
Everyone who uses AI is familiar with the so-called “garbage in, garbage out” principle. If the inputs are weak, biased, incomplete, or poorly contextualized, the outputs will be too, hence requiring further verification.
In impact investing, decisions often depend on local context, qualitative information, lived experience, and imperfect data. If AI tools draw data from unverified sources, they may over-weight the most available or popular information, or reinforce existing biases.
While AI can also make it easier to summarize papers, generate memos, and move quickly through information, it can also make us overly reliant on such tools and lose the ability to engage with complex materials critically and develop our own analytical judgment. This is an important consideration for students and early-career professionals who are still developing critical thinking and professional judgment, but it is equally relevant for experienced practitioners as AI increasingly shapes how we access, interpret, and engage with information.
The poll results reflected that the greatest concern among practitioners was the potential loss of human judgment, followed by data bias, quality and cybersecurity.
So, should (or can) AI be avoided? Probably not. The more important question is how accountability and governance can be strengthened around the final outputs shared with clients, investment committees, partners, and communities.
Equity and emerging market contexts require special care
A major question is how AI will affect work in emerging markets, frontier markets, and underserved communities.
AI tools generally perform best where data is abundant, structured, and easily accessible. But many impact contexts are precisely the opposite: data may be limited, uneven, externally produced, or missing altogether. At the margins, important realities can be ignored because they are not well represented in the data.
This represents both risks and promising practices. On the risk side, AI may reinforce existing visibility gaps by privileging what is already documented, digitized, and popular. This could disadvantage smaller organizations, less formal markets, Indigenous/grassroots actors, or fund managers with fewer public materials.
On the other side, AI tools may be beneficial within localized and closed data systems, including community-specific databases, internal knowledge repositories, and curated datasets. AI can also aid in data collection, including using interviews or field-level inputs to better understand local challenges and intervention outcomes.
AI tools should not simply import external assumptions into local contexts, but be carefully shaped by the people and communities most affected by the decisions being made.
AI should be treated as infrastructure, not just a product
Baldwin invited participants to think of AI not only as a product purchased from large technology companies, but as infrastructure that organizations and communities help shape.
This framing changes the role of practitioners. Rather than being passive users, impact investors and development finance professionals can act as stewards, asking what kind of systems they want to build, whose values are embedded, and who gets to participate in shaping the future of these tools.
In this framing, AI adoption is not only a technical issue. It is also a governance issue, a data sovereignty issue, an equity issue, and a learning issue. For the impact investing and development finance sector, this creates an opportunity at the intersection of capital, accountability, social outcomes, and long-term systems change.
What the sector needs next
AI is already helping impact investing and development finance teams move faster, synthesize more information, and experiment with new forms of analysis and communication.
But the deeper question is whether the sector can use these tools in ways that strengthen judgment, protect trust, improve decision-making, and remain accountable to the communities and outcomes that impact capital is intended to serve.
Need #1: Organizations need practical guidance on AI use
This includes organizational policies, data management practices, cybersecurity safeguards, clear internal norms, and decision rules for when AI is appropriate (and when it is not). Organizations also need clarity on who is responsible for reviewing AI-generated work and how outputs should be verified before they influence investment, impact, or community-level decisions.
Need #2: Human oversight must remain built into AI workflows
AI should support professional judgment rather than replace it. Organizations need clear review processes, accountability for final outputs, and safeguards against relying on AI-generated analysis without sufficient verification or contextual understanding.
Need #3: The sector needs more spaces for shared learning
Practitioners are experimenting quickly, but adoption remains uneven. Some organizations are building sophisticated workflows, while others are still developing basic policies and testing individual tools.
Rather than using a one-size-fits-all approach, these conversations should focus on shared principles that are organization- and/or location-specific. Such principles may include transparency, accountability, privacy, inclusion, contextual awareness, human oversight, and a commitment to strong AI governance.
Many of the insights reflected the practical experiences of organizations actively applying AI in investment, impact measurement, and development finance workflows. The CAFIID community is committed to creating shared spaces for honest exchange about what is working, what is not, and what still needs to be tested.