How AI Paints a Deeper Market Picture With Untapped Knowledge

MSD Hub editor's note (Michael Field, Senior Systems Specialist, Vikāra Institute):

The Author highlights the increasing importance of leveraging emerging AI tools to complement and improve the application of systems thinking approaches. Of particular importance is a recognition that complex social, political, and market contexts are dynamic requiring ongoing probing and learning processes. As a result, the author points out that AI is best utilized when perceived as a tool that learns with the team, taking on the time-consuming tasks of collating and pattern sensing various sources of data that teams can then use to interrogate and filter through human lenses and perspectives. So, while AI is a powerful tool, it needs to be understood that it is only reflecting the facts, assumptions, biases, and myths masquerading as facts that are interwoven into human knowledge sources. Good systems thinking approaches still require a probe, learn, and adapt process to effectively catalyzing positive change, and AI tools have huge potential to supercharge such learning and doing processes.  

---

Bridging the Knowledge Gap

Despite robust data and indicators, management at MSD organizations often finds itself asking: “Why did this outcome occur (or not) in the market?” 

Even as this question is posed, some pockets of the organization, typically local field teams, already know the answer. This tacit knowledge, defined as the insights in people’s heads, helps make sense of local context, relationship dynamics, and norms or behaviors. It adds rich context to reveal the “story behind the numbers,” which metrics and indicators alone simply cannot capture. This empowers MSD teams to adapt programs more intentionally, and thereby increase their chance of achieving true impact at scale.

Until recently, extracting and sharing this critical knowledge across an organization proved a mammoth task. Usually, it was captured informally, in unstructured forms; and then stored in scattered locations, from central repositories to private folders. 

But now with recent advances in AI, tacit knowledge in all its messy formats can be consolidated for analysis in secure platforms, enabling MSD teams to make more informed decisions. In this blog, we explore how certain AI platforms aggregate a wide range of data quickly to assist teams to better understand, and better support, markets in which they work.

When Numbers Are Clear, but the Story Is Not

Consider a composite MSD example: a mid-sized organization launches a program in Rio de Janeiro to expand household purchases of heat reflective roof coating among low-income communities (to lower indoor temperatures). While the program unlocked supply bottlenecks and raised household awareness, monitoring indicated it did not translate to a higher uptake of reflective coating products.

Local field staff from the organization, however, knew the reason. They observed that in many low-income homes, reflective coating proved ineffective because no ceiling separated the roof and living spaces. Early adopters also reported dust, dirt, and mold reduced effectiveness over time, with word-of-mouth dissuading neighbors. 

Perhaps most importantly, local field staff recognized the role of social norms. New tin roofs signaled rising status in one’s community, so painting over them (even reflectively) was viewed as downgrading an aspirational marker. But as often occurs in MSD programs, this recognition, while extremely powerful, was not systematically captured nor acted upon.

This example shows how intentionally documenting tacit knowledge adds essential context to traditional Monitoring and Evaluation (M&E) metrics that would inform timely program adaptations (e.g., reframing reflective coatings as a status-enhancing upgrade). Looking ahead, organizations that effectively merge valuable, frontline insights with quantitative data will uncover deeper market system behaviors, and unlock greater impact in the process.

How AI Supports Sense-Making in Complex Systems

Historically, deriving meaning from large volumes of unstructured tacit knowledge has been time-consuming, if not altogether impractical. But now, recent advances in AI make it possible to synthesize what teams know, in near real time.

Using qualitative AI platforms, organizations can analyze diverse, often incongruous sources in a secure workspace. These platforms aggregate knowledge products spanning M&E indicators, household survey data, staff field notes, weekly team meeting transcripts, and more.

To ensure high-quality insights feed into the AI platform, an M&E lead or knowledge manager should conduct light to moderate data validation, and where needed, engage directly with practitioners to clarify or deepen raw inputs. Through this effort, a more detailed market picture will take shape.

At the same time, tacit knowledge rarely surfaces through organic documentation alone. More often, it emerges from complementary learning practices. A few examples are outlined below, with the caveat that these approaches should always be tailored to team culture and capacity:

• Learning sessions, recorded and transcribed for AI, to unpack deeper observations (e.g., mid-year reviews and "pause-and-reflect" discussions).

• Targeted interviews to uncover market drivers (one-on-one or small-group).

• Micro-reflections (approx. 10 minutes) to build a cumulative market picture over time (e.g., weekly voice notes or written responses). 

• Debriefs with market actors or experts to broaden the systems view with external perspectives (e.g., stakeholder interviews).

Once data has been verified and added to a qualitative AI platform, the sense-making begins. Many AI tools allow users to engage directly with their data. For instance, they may ask why an indicator is higher or lower than expected, and then follow up with questions to draw upon all the unstructured information uploaded for analysis. Returning to our example, if uptake for reflective coatings is low, management can query an AI-powered database for a range of drivers, to identify hidden trends. 

And as a final step, findings from any AI tool should always be validated with local field staff to confirm their accuracy. 

A Few Notes of Caution When Using AI

In complex market systems, AI is best seen as a learning partner. It quickly identifies patterns, but depends on staff to frame and interpret them. All this to say, MSD practitioners must apply their own expertise to shape prompts and prioritize frameworks (e.g., directing AI to focus on specific MSD frameworks rather than generic reasoning models).

Additionally, it is essential to opt for enterprise-grade AI platforms that apply strict security standards. Such platforms will keep inputs from training external language models, and ensure an organization’s information is safeguarded.

Putting It All Together: Implications for MSD Practice

Most MSD organizations already hold valuable, untapped data across their programs. They simply need light-touch AI tools to help them to nimbly connect the dots. 

For those MSD practitioners seeking to integrate AI into MSD learning cycles, several principles should be kept in mind:

• Rapidly uploading tacit knowledge into AI platforms ensures they remain timely for adaptive management.

• For a deeper systems-wide view to emerge, teams need clear signals from leadership that documenting challenges and pain points is an act of learning, not failure.

• External constraints like donor expectations, political context, and partnership dynamics will inevitably influence how insights translate into action.

By combining traditional M&E metrics with practitioner knowledge in AI-powered tools, MSD teams can transform individual experiences into a powerful, shared evidence base. While both qualitative and quantitative data have their biases, together they provide a more complete picture of how systems behave. And this richer knowledge base, in turn, makes adaptation easier and more effective.

Author: Monica Rashkin, Founder and CEO, Bright Insights