How Avante turned scattered information streams into prioritized intelligence with Consili.
Augmenting analysis with AI
Information analysis drives progress in modern organizations. With the help of AI, we're accelerating a growing share of this work.
In the digital age, access to information is no longer the problem; it’s the sheer volume of it that hinders momentum. For sectors like law, insurance, and compliance, the daily influx of regulatory updates, news feeds, and analysis creates a state of paralysis. This "Information Overload" forces highly skilled experts to spend hours filtering noise instead of analyzing signal.
We set out to test whether generative AI and specifically large language models could play an effective role augmenting a human analyst workforce. Specifically, can AI agents be leveraged to perform analytic tasks and free up human capital for more valuable pursuits? To answer this question we needed to test whether it was even possible to deploy a reliable AI analyst and then measure the productivity gains relative to the investment. We built Consili to answer these questions.
Transforming data noise into signal
The value of Consili became clear when legal and public relations professionals used it to monitor their noisiest data feeds, allowing them to uncover insights faster and with greater accuracy.
We designed Consili as an agentic knowledge discovery platform to serve the needs of knowledge workers. It offers an easy-to-use platform exposing the full capabilities of AI agents to users with no technical background. Users can effortlessly define their research areas of interest, subscribe to data sources and have Consili automatically monitor them and generate relevant insights.
How Consili works
At one end, Consili ingests the information sources that users specify. At the other end, expert analysts work from Consili Insights.
Consili architecture
- Crawler: Pulls in the latest data from the user specified sources.
- Interests. Set up a knowledge discovery campaign by defining interests and connecting them to information sources.
- Analyst: The AI synthesizes key information from every source to populate user feeds with Consili Insights.
- Assistant: Choose an Insight then chat with Consili about it. The assistant backs up its claims with references to the source material.
- Feedback: As users interact with Consili, it uses each interaction to refine its algorithms and improve future Insights.
Information triggers
One capability that emerged from user feedback was the need for automated responses. We built Triggers to initiate specific actions when Insights meet predefined criteria—enabling everything from automated alerts to API-driven workflows. Where oversight is required, we included human review gates.
This proved especially valuable for PR teams monitoring brand mentions and legal teams tracking regulatory changes, where speed of response is critical.
Consili Insights
Consili does the tedious part, generating Consili Insights for knowledge workers. An Insight is a summary of a relevant piece of knowledge, based on interests that users specify, and always backed with references to the source material. Consili understands business context and who the stakeholders are. It's fluent in the jargon, concepts, and language of diverse industries.
Insights are only the starting point. Consili frees experts from the tedious part of the job and gives them tools for building a deep, clear understanding of what's happening.
Hallucination Management & Continuous Learning
All generative AI is prone to hallucinations. Rather than trying to prevent them entirely, we engineered a system to manage them effectively:
Consili does this in a couple of ways:
- Always-on source attribution. Consili backs every Insight with references to source material. Users are empowered to validate sources with a click to see for themselves.
- Continuous feedback loops. As Consili learns more about each business and industry, it is able to generate increasingly accurate Insights.
- Negative Insights: We added the ability to flag intentional false negatives—things that should have been identified as relevant but weren't. This, combined with user feedback, improves future analysis.
Consili evolves with changes in organizations, progressively learning and adapting. Like the dynamism of human cognitive growth, Consili improves its responses to new information and perspectives.
Consili's in-built feedback mechanism fosters a two-way interaction with its users. This dynamic not only amplifies the user experience but refines the system's analytical proficiency over time.
Knowledge Discovery for Legal & PR Teams
Traditional knowledge discovery tools like Google Alerts are a starting point. We built Consili to keep climbing—monitoring multiple streams simultaneously, filtering signal from noise, then enabling deep exploration through conversational AI.
What this enabled:
- PR teams stayed on top of brand mentions across news, social, and blogs
- Legal teams tracked regulatory changes without manual daily checks
- Analysts explored scenarios and identified risks through chat interface
Information Synthesis for Research Teams
Every knowledge worker's job starts with information overload. Consili takes large volumes of unstructured information and tailors synthesis to unique priorities and interests—always with source references. The system understands audience needs too, generating different formats for different stakeholders.
What this enabled:
- 80% reduction in time spent on initial information gathering
- Consistent format across team members' research outputs
- Ability to customize insight depth based on urgency
Building and deploying Consili provided critical insights:
- RAG architecture with source attribution is non-negotiable for enterprise AI in high-stakes environments
- Continuous feedback loops aren't optional—they're how AI systems become genuinely useful
- Custom triggers and automation were the most requested features, showing that passive insights aren't enough
- Industry-specific customization (terminology, priorities, workflows) drives adoption more than general capabilities
The Challenge
With the rapid acceleration of AI, many firms struggle to find reliable approaches to its effective market deployment. This makes it essential to establish production grade solutions that have been hardened through real world use.
Our Solution
We built Consili to bridge that gap. It is a reference architecture that formalizes production-ready methods for using AI in high-stakes environments.
It encodes our hard-won lessons on:
- Reliable Ingestion: Handling diverse data sources at scale.
- Hallucination Management: Solving accuracy through architectural design, not disclaimers.
- Measurable Accuracy: Feedback loops that improve the system over time.
- Oversight: Integrating automation without sacrificing human control.
What This Means for Clients
Working with Avante, is partnering with a team that's been there. Clients benefit from leveraging an architecture proven in production, adapted to their specific domain.
- Rapid Prototyping: Apply proven patterns to your use case to show value in weeks.
- Iterative Delivery: Deploy incrementally based on real usage, not guessed requirements.
- Production from Day One: Skip the prototype-to-production gap entirely.
🇨🇦 Built in Toronto