Introducing auto-mode for Claygent. It picks the AI model for each row of research, balancing answer quality AND cost. Now on by default for new Claygent columns: - 47% cheaper than sending every t
The Hook
“Today, we're launching Claygent's Auto Mode, an easy way to route the right task to the right model for Claygent.”
Problem Agitation
“We have a lot of models in the Claygent model picker, and it's hard as a customer to know when to use what model and what model will actually get them the best results...”
Product Reveal
“Some are deep research-like tasks that are very involved and long-horizon, and there are some that are more contact lookups... effectively calibrate the model... Pareto frontier of cost performance...”
Feature Teaser
“So our framework for when to do post-training is typically around what kind of alpha do you have inside of your company that you can operationalize into models...”
Call to Action
“Claygent Auto Mode is available today. We're really excited for you all to try it.”
Synthesizable in Remotion · key components of the ai-feature-fireside-explainer template.
Draws an X/Y axis line chart with plotting data points, dashed curve stroke, and an emphasized callout pill at the optimal frontier coordinate.
Use interpolate on frame to stroke SVG paths. Animate dots using spring opacity and scale, followed by an Auto Mode tag badge fading in at the inflection point.
Simulates a spreadsheet data view where rows dynamically populate model assignments like 'Frontier model' or 'Fast model' based on task complexity.
Map rows with staggered delay. Render column pills with conditional color accents and pulsing badge states.
Horizontal stacked bar chart showing volume shifting across three time intervals among frontier, open-source, and custom models.
Render three row containers. Animate width of segmented flex items using spring physics and distinct hex fills, accompanied by synchronized legend items.
Purpose. Establish feature utility and high scale validation (5B+ runs).
Execution. Host direct address supported by a clean stat card overlay.
Purpose. Clarify cognitive fatigue caused by choosing between dozens of fast-moving LLMs.
Execution. Screen mockup showing an overflowing model selector dropdown.
Purpose. Build trust by disclosing calibration methodologies, benchmark trade-offs, and router design.
Execution. Fireside discussion cut alongside bespoke 2D technical motion graphics.
Purpose. Drive immediate feature adoption.
Execution. High-fidelity feature card over blue brand gradient.