tl;dr, just show me the product:

• For a 3m product demo, see this youtube video

• For an extended 50m AI Financial Analytics Speed Run / Tutorial, see this youtube video

• To signup and play with a demo agent, follow this link

After having been in beta for > 6 months with a text-to-SQL product, we’re releasing an expanded version of Patterns that implements an agentic system and elevates the product experience for non-technical business users.

Our business focus is in three key areas:

1. Non-technical end-users

2. Database analytics

3. A hybrid service offering

Non-Technical End-Users Need Autonomous Agents

and why we’re different than a BI dashboard with AI

Initially, we started with a simple text-to-SQL product, but through months of experimentation, we realized that competing with bolt-on AI features for BI tools required a deeper approach; and people hate dashboards. So to truly win, we needed to go beyond generating SQL and simulate the entire experience of working with an analyst. For a product aimed at non-technical users, it's unacceptable to just render SQL or, worse, a SQL error. Our approach involves generating responses to user requests that:

• Clarify the user's questions and understand their intent

• Plan an entire analysis, which might involve multiple SQL queries

• Generate SQL

• Iterate on SQL, correcting for errors and empty results

• Evaluate query results and cross-check against prior results

• Generate charts and visualizations

• Iteratively improve chart interpretability

• Interpret, analyze, and generate insights from query results and charts

• Compile a final report based on all queries, charts, and insights

Expanding our product scope to include these capabilities elaborates on the real-world product differences between an AI-first product and AI features. We no longer see the need for a "dashboard" in the traditional sense when you can just get an answer on demand by an agent.

It’s a distinction, but an important one. Is the AI helping a technical person create a dashboard? So that they can create more dashboards faster for their team. Or is the technical person creating an agent that can interface directly with end-users and serve requests directly?

The former maybe makes the analyst 30% faster, the latter unlocks data access to everyone driving 2-10x more value to the business.

Database Analytics

Similar products, like ChatGPT, handle CSV uploads and Python interpreters well. While these features are cool and easy to bring to market, they don't align with our ideal customer profile. If a company doesn't have a database, it doesn't have a significant data problem. We’re building for companies where data teams are overwhelmed, have hundreds of tables in their data warehouse, and Jira boards full of unanswered data questions.

Building an LLM-to-Database interface (and an enterprise workflow tool to boot) is a complex task that includes database inspection, semantic understanding, and the generation of an enterprise knowledge graph to support query generation. All of this information is neatly organized and added to our prompt engine that calls relevant context when needed - giving the AI the info it needs and nothing more.

Designing for 100s of tables, underwater data teams, and non-technical users, places us on an entirely different product vector than most AI-powered data tools in the market and existing BI tools.

AI as a Proxy and Hybrid Services

Excuse the marketing lingo, but I genuinely believe we're early in the trend of hybrid services—human and AI-powered services that encapsulate an entire business function.

Task → Workflow → Functions

• Task automation was SaaS 1.0, with simple CRUD apps executing actions like adding a lead to Salesforce.

• Workflow automation is current-day SaaS, where integrated data and basic ML automate entire workflows, like scoring a new lead and sending an email.

• Functional automation is where an entire job description can be fulfilled by software, such as an AI taking over the role of an SDR or Data Analyst.

We operate in teams because a single individual can’t deliver an entire function. A data team typically has a manager, engineer, scientist, and analyst, each with distinct workflows and tasks powering the "data team."

Data teams scale unprofitably when many stakeholders have numerous data requests, overwhelming the single analyst who serves the BI function of the company. Patterns is designed to address this—ensuring you need no more than one analyst on your data team. Patterns provides that analyst with a clone of themselves, imbued with the organization's knowledge, acting as an AI twin or a proxy. This AI proxy performs work on behalf of the human it represents, only contacting them when it encounters an issue or needs further information. This enables analytics teams to scale beyond their human capacity.

This not only opens opportunities for data teams but also creates a massive opportunity for consultants and freelancers to deliver value at scale. Imagine being an independent consultant supporting 10 companies through your AI bot, handling issues as they arise. It’s like having a grad student—one that significantly enhances your capacity to deliver value.

Add Patterns to Your Data Team, or Hire Us as Your Data Team

While Patterns is designed for non-technical users, is requires some technical involvement to be successful. A technical person must:

• Setup: A clean data model, established by a technical team member, is crucial for enabling the AI to accurately access your database. If a human couldn’t onboard without help, our AI won’t be able to either... yet.

• AI Training: Iteratively, by asking questions and evaluating responses, you must integrate business context like metrics and definitions, usually done by someone technical.

• Ongoing Support: Continuous updates and optimizations need a technical "human-in-the-loop" to maintain system performance.

And so we’ve positioned our product/service offering around that:

I’ve got someone technical - $1,000/month

1. Unlimited Users, Agents, Requests

2. 30-day money-back guarantee

3. Onboard with our Agentic Analytics implementation and management guide

Take care of it for us - $2,500/month

1. Unlimited Users, Agents. Requests

2. Whiteglove onboarding, data cleaning, modeling, and prompt engineering

3. Human-in-the-loop support

4. 30-day money-back guarantee