Comparison Table
| Problem 1 | Problem 2 | Problem 3 | |
| The Problem | Cross-functional alignment: same words, different meanings; approvals are hard to get | Hidden dependencies & unknown unknowns in regards to legal challenges & implications | AI pulls teams into solution mode and away from customer discovery |
| Evidence Strength | Strong. Raised by 8 PMs. It also came up across company types, so it isn’t specific to one industry. | Moderate. Raised by 4 PMs. It was described in vivid detail, but mostly in regulated or public-company contexts. | Weak. Raised by 3 PMs, plus 2 soft mentions. It came up more as a concern than as a daily pain point. |
| Current Workarounds | Slack and Doc comments, concept-review meetings, agendas with a “this meeting succeeds if X” line, and writing proposals down so people can react. In practice, the workaround is the alignment meeting itself plus PM follow-ups. | Bringing legal in early, roadmap review sessions, and compliance gates added after being burned before (e.g., small-business lending). These are mostly reactive. | Discipline only: staying in the problem space and insisting on real user contact. No interviewee named a tool that addresses this. |
| Competitive Gap | The space is crowded on visibility: Productboard, Jira PD, Notion, Slack, approval workflows, and glossaries. The irony is that this abundance creates its own problem. Decisions, comments, and context end up scattered across platforms, so no one has the full picture, which deepens the misalignment these tools are meant to solve (echoed in long list #6 and #8). The gap is that existing tools give everyone access to the same artifacts but never check whether people interpret them the same way. | Jira tracks dependencies only once they’re known. Nothing surfaces legal or cross-team implications during planning. Compliance tools exist, but they live outside the PM workflow. | The AI research tool space is saturated (Dovetail, HeyMarvin, Maze, Sprig), but these tools speed up synthesis of research teams already did. None of them push teams back toward customers, and some may accelerate the shortcut. This makes a crowded adjacent market with very little direct whitespace to sell into. |
| ICP | PMs who coordinate across eng, design, legal, and research teams that each use different vocabulary and communication styles. This is broad, so it would need narrowing. | Fintech and regulated-industry PMs, especially at public companies. This is narrow and sharp. | Discovery-heavy PMs on teams adopting AI quickly. A very vague segment. |
| Market Size | The fintech SAM ($7.5M) works as a starting idea of the market. The true ceiling is much larger, since the problem isn’t fintech-specific. | The fintech $7.5M SAM / $225K SOM maps directly onto this ICP | Small and hard to monetize. Teams rarely pay to slow themselves down. No estimate yet. |
| Buildability | Medium. There are clear narrow entry points: an “interpretation check” where stakeholders restate a proposal and AI flags where their versions diverge, cross-platform comment synthesis, and clarifying what an approval actually covers. It would almost certainly need Slack and Docs integrations. | Hard. It requires regulatory domain knowledge and company-specific context. It also carries liability risk if it misses something, since an “AI lawyer” that’s wrong is worse than none. | Medium. Prompts and nudges are simple to build, but the fix is behavioral, so measuring impact is tough. |
| AI fit | Strong. Comparing language, spotting ambiguity, and synthesizing across sources are core LLM strengths. Since 5 of 9 PMs re-verify AI output, it should keep a human in the loop. | Moderate. Pattern-matching for regulatory flags is plausible, but the stakes of hallucination are highest here, and PMs already distrust AI. | Weak. AI is the cause, so adding more AI risks deepening the problem. At best it fits as a guardrail. |
We are leaning toward problem 1: Cross-functional alignment: same words, different meanings; approvals are hard to get.
Across our interviews, nearly every PM brought up some form of communication or alignment challenge, although the specific pain points vary. Some described the difficulty of keeping different teams aligned throughout a project’s development, while others pointed to more concrete workflow issues, such as comments scattered across multiple platforms and the constant back-and-forth between Slack, Google Docs, and meetings. These existing workarounds suggest that PMs are already investing considerable time and effort into addressing the problem, yet their current tools and workflows do not seem to fully resolve it.
We chose this problem because it is both widely experienced and important to the PMs we interviewed. Some interviewees also suggested that emerging AI tools could potentially improve cross-functional communication. More importantly, the problem offers several concrete and potentially buildable entry points where AI could add value beyond existing collaboration tools. Rather than attempting to solve cross-functional alignment as a whole, we could begin with a narrower use case, such as translating technical terminology into context-appropriate language for nontechnical stakeholders, synthesizing fragmented comments across different platforms, or identifying discrepancies in how different teams interpret the same project requirements. These possibilities also give us a way to explore whether AI can reduce the manual effort involved in communication rather than simply introducing another tool into an already fragmented workflow.
Our preliminary market sizing also suggests a potential business opportunity. Focusing initially on fintech companies with frequent cross-functional collaboration needs, we estimate a serviceable available market (SAM) of approximately $7.5 million annually, based on 12,500 potential paid users. Assuming we capture 3% of this market within three years, our estimated serviceable obtainable market (SOM) would be $225,000 in annual recurring revenue, representing approximately 75 companies and 375 paid users. While these estimates rely on assumptions that still need validation, they provide an initial indication of the market opportunity and a starting point for refining our ICP.
One idea we took from this week’s reading, How to Identify Your Ideal Customer Profile, is the importance of starting with a narrowly defined customer segment rather than trying to serve everyone who experiences a problem. We realized that although cross-functional misalignment appears common among PMs, we still need to define our ICP more precisely. For example, we could narrow our focus to PMs in fintech, companies of a particular size, or PMs who regularly coordinate across multiple teams with different communication styles and tools. We could also start with a smaller, more specific problem, such as context-aware translation of technical terminology or cross-platform comment synthesis. Our next step is to determine which customer segment experiences the greatest challenge regarding which part of the communication.
