A strong product strategy should always start with the user problem and pain point, not with the technology itself - Garima Golchha, Clarity AI

Garima Golchha is a Product Manager at Clarity AI, a sustainable tech company that leverages advanced technology and AI to bring societal impact into financial markets by giving investors the data and tools that help redirect capital toward more sustainable activities. 

Her role is to work with clients and understand their needs, such as - especially for Article 8 or 9 investors, sustainability reporting requirements. She then turns the answers to these needs into product features. Product thinking is a capability startups need “often from the beginning” she explains:

“PMs can have the biggest impact in shaping direction, validating product–market fit, and deciding what not to build. 

Of course, that role does not necessarily need to be labeled as a “PM” role. Someone (often the CEO, tech or strategy), could end up doing product discovery in the early days. But those early product decisions are critical in shaping a product that can scale sustainably over time.” 

‍

At Clarity AI, she uses AI to “simplify complexity”, turning  “dense regulations and fragmented sustainability data and turning them into workflows that investors can easily understand, trust, and use.”

For other impact startups that work with AI, Garima says that it’s a technology that can bring both significant opportunities and critical risks:

“I usually think about opportunities in two areas: simplifying workflows externally and internally.

Simplifying client workflows: In my case, we use AI across different parts of the data and reporting process. This includes NLP models for data collection, machine learning models to estimate data gaps, and generative AI capabilities to help clients report faster and navigate complex sustainability requirements with more confidence. More broadly, AI can compress hours of analyst work into minutes.

Simplifying internal workflows: AI can remove friction inside teams. For example, within Product, this can mean using AI to generate and maintain PRDs and documentation, explain code and data pipelines, and create internal tools (like GPTs and NotebookLMs) that allow non-technical teams to ask questions and get them resolved without relying on a few experts.

In both cases, the goal is to automate mundane work and free up time for more strategic decisions that actually add value.”

‍

On the other hand, the biggest risks she sees are:

“Hallucinations and lack of explainability: Especially in highly regulated environments, users want data that they can trust and trace back to clear sources and assumptions. If outputs cannot be explained or justified, they quickly lose value, regardless of how advanced the AI is.

Misaligned incentives: Teams sometimes optimize for “wow demos” when users actually need consistency and correctness. Not everything needs flashy AI. Sometimes, “boringly correct” is the real differentiator.”

‍

As a Product Manager using AI as a core technology in her work, Garima emphasizes that with all the AI-enhanced tools existing today, the main challenge is no longer capability.

“AI is a means, not the goal. A strong product strategy should always start with the user problem and pain point, not with the technology itself.

From there, the key is judgment: being clear about where AI genuinely adds value and where it doesn’t. In some cases, a deterministic or rule-based solution works better, and choosing that is often the right product decision.”

‍

She also explains that AI’s goal should only be to improve the product and not add more useless complexity:

“A principle I strongly believe in is:
The best AI products won’t feel like AI. They’ll feel like the shortest path between a question and a defensible answer.

AI should reduce cognitive load, not introduce a new one. If users need to “figure out” the AI, something has gone wrong.

The key here is being relentlessly use-case driven. Again, AI is a means, not the goal. Focus on problems and pain points, not the solution. Talk to users, observe how they use your product, and understand what they are actually trying to achieve. Build products that fill that gap, whether AI is involved or not.”

‍

Then, there’s also a matter of building trust in an AI product, not just developing and launching the new solution. Garima adds that everything starts with explainability - helping people understand how things work. 

“At Clarity AI, we use AI in areas like data collection, estimation, and generative AI insights and reporting capabilities. Building trust there means:

• grounding estimations in high-quality data and robust methodologies
• combining multiple sources, such as company disclosures and internal data science models
• making any assumptions reasonable, transparent, and clearly labeled.”

‍

All of this has to be inline with multiple stakeholders’ expectations - from clients, to the commercial team, engineers, research and leadership.

“For example, a classic tension, especially in startups, is speed versus robustness. The business wants to move fast to be first to market, while the tech team worries (rightly) about introducing shortcuts (“ñapas,” as they say in Spanish) and long-term maintainability. 

If you decide to move fast, you do so consciously, often through scoped MVPs or early access releases, and you explicitly plan time to address the resulting tech debt. The goal isn’t to avoid trade-offs, but to make them explicit, shared, and intentional.”

‍

Finally, we wanted to know how Garima sees the role of Product Managers who work with AI products evolving in the next 3-5 years:  

“I think many PM tasks like analysis, reporting, and documentation will increasingly be supported or automated by AI. However, the core PM skills - communication, motivating teams, and building the right thing - will always matter.”

Thank you, Garima Golchha!