Inspiration
Businesses make important decisions based on assumptions.
"This is our best market." "Our customers want this." "This competitor is our biggest threat." "This investment will create the most growth."
The problem is that these assumptions are often built from incomplete, fragmented, or outdated information.
Most AI assistants will take the user's premise and help them act on it.
We wanted to build something different:
What if AI could investigate the premise itself?
That became Business Intent — an AI-powered strategic intelligence system that investigates business assumptions against real-world evidence, searches for contradictory signals, and turns those findings into actionable decisions.
What it does
A user gives Business Intent a company, strategic question, or business assumption.
For example:
"Is expanding into Europe actually our highest-leverage growth opportunity?"
Instead of immediately generating a recommendation, Business Intent investigates the question.
The system breaks the strategic problem into smaller research tasks, gathers relevant information, evaluates supporting and contradicting evidence, and synthesizes the findings into a strategic reality check.
The output answers questions such as:
- What evidence supports the assumption?
- What evidence contradicts it?
- What important signals are we missing?
- What alternative explanations exist?
- What should the team investigate or do next?
The core workflow is:
Question → Investigation → Evidence → Contradictions → Synthesis → Action
The key difference
Business Intent is not designed to simply confirm what the user already believes.
It deliberately looks for information that could challenge the original hypothesis.
For example, if a company believes:
"Enterprise customers are our best growth opportunity."
Business Intent can investigate the assumption and compare signals across markets, competitors, customers, industry developments, and other available sources.
It may find evidence supporting the hypothesis.
But if it also finds evidence suggesting another segment has stronger demand, lower acquisition costs, or less competitive pressure, that contradiction becomes part of the final analysis.
The system therefore treats:
Supporting evidence + contradicting evidence
as equally important inputs to the decision.
How AI is used
AI is central to the product rather than being an interface layer.
The system uses AI to:
- Interpret an ambiguous strategic question.
- Decompose it into researchable hypotheses.
- Generate and prioritize investigation tasks.
- Analyze information gathered from different sources.
- Compare evidence against competing hypotheses.
- Identify contradictions and missing information.
- Synthesize multiple signals into a strategic conclusion.
- Recommend what the user should investigate or do next.
This creates an agentic workflow rather than a single prompt-response interaction.
How we built it
Business Intent is structured as an agentic intelligence pipeline:
Strategic Question → Intent / Hypothesis Decomposition → Research Tasks → Evidence Collection → Evidence Analysis → Contradiction Detection → Synthesis → Recommendation
The separation between research and reasoning is important.
Rather than asking an LLM to generate a strategic answer directly, the system first creates an evidence base and then reasons over that evidence.
This makes the application more useful for questions where the answer depends on information that changes over time.
Example workflow
A user asks:
"Should our company expand into Europe?"
Business Intent can transform that broad question into multiple investigations.
For example:
Market
- Is demand growing?
- Which segments are growing fastest?
Competition
- Which competitors are expanding?
- How crowded is the market?
Customers
- Are there signals of demand?
- Which customer segments are most attractive?
Risk
- What regulatory or operational factors could affect the decision?
The system then combines these investigations and looks for relationships and contradictions between them.
Instead of returning a generic market report, it produces a decision-oriented analysis.
What we learned
Our biggest lesson was that building a useful AI product requires much more than connecting an LLM to an interface.
The difficult engineering problem is designing the workflow around the model.
We had to think about:
- How to decompose ambiguous questions.
- How to keep research focused on the original objective.
- How to distinguish evidence from assumptions.
- How to handle conflicting information.
- How to reduce confirmation bias.
- How to transform research into an actionable output.
This led us to think of AI agents less as chatbots and more as systems that can investigate a problem from multiple directions.
Challenges
Real-world information is messy.
Relevant information can be distributed across many sources, change over time, and contradict itself.
A naïve AI research system can also suffer from confirmation bias by searching for information that supports the user's original assumption.
We therefore focused on making counter-evidence a first-class part of the workflow.
The system is designed to ask not only:
"What supports this hypothesis?"
but also:
"What would make this hypothesis wrong?"
Impact
Business decisions are expensive.
A company can spend months and significant resources executing a strategy that was based on an incorrect assumption.
Business Intent is designed to provide an additional layer of intelligence before that commitment happens.
Instead of only helping teams execute decisions faster, it helps them investigate whether the decision itself is supported by the available evidence.
The broader goal is to make strategic decision-making more evidence-driven, less dependent on confirmation bias, and more adaptive to changing conditions.
What we built
Business Intent is a working prototype of an AI strategic intelligence system that connects:
Strategic Questions → Agentic Research → Evidence → Contradiction Detection → Recommendations
The current prototype demonstrates how AI can move beyond answering questions and instead investigate the assumptions behind them.
Future vision
The long-term vision is a continuously updating intelligence layer for organizations.
Instead of waiting for someone to ask a question, Business Intent could monitor changes across markets, competitors, customers, economics, and company strategy.
When new evidence conflicts with an important business assumption, the system could proactively surface it.
The goal is simple:
Don't just use AI to decide faster. Use AI to ensure you're making decisions based on a better understanding of reality.
Built With
- ai-agents
- data-analysis
- docker
- fastapi
- github
- google-cloud
- google-gemini
- javascript
- llms
- natural-language-processing
- python
- react
- rest-apis
- typescript
- web-research/data-apis
Log in or sign up for Devpost to join the conversation.