How Mutiny Uses Claude to Give Every Sales Rep a Full Creative Team
Mutiny, a GTM automation startup, rebuilt its entire platform around Claude after Anthropic’s Opus outperformed competitors on its internal design evaluation benchmark. The agent-first architecture lets sales reps describe what they need and Claude generates fully branded assets in a single shot from just a website URL. Design satisfaction scores improved 3x after making Claude Opus the default model, and asset creation is now 4.5x faster than previous workflows.
Impact
3x
Improvement in design satisfaction
4.5x
Faster asset creation for sales teams
9 out of 10
Sales reps reporting competitive edge
Challenge
Early LLM integrations were siloed — each AI capability required narrow guardrails and could not be combined dynamically, preventing Mutiny from delivering a fully autonomous, multimodal creative agent.
Solution
Mutiny re-architected its platform around Claude Opus as the default model, building an agent-first experience with tool-based access for research, brand building, and design. An LLM-native data model aligned to frameworks Claude performs well on enables fully branded asset generation in a single shot.
Tools & Technologies
What Leaders Say
“We really saw an opportunity to re-architect the whole system and make this a very agent-first experience. The agent is now multimodal and has tool-based access to do research, build the brand, and design the entire experience.”
“It was the highest benchmark we had achieved on our internal design eval. Not only are we seeing high-quality outputs, but a much wider range of those outputs as well.”
Sign up to read complete case studies, access detailed metrics, and unlock all use cases.
Full Story
Mutiny builds tools that help GTM teams create the assets they need to generate pipeline and close deals — personalised pitch decks, deal rooms, and business cases generated in the customer’s own brand. The vision was always to eliminate the bottleneck of waiting on marketing or design.
For years, that vision was constrained. Early LLM integrations were siloed: AI could pull brand styling, generate text variations, or conduct research, but each task required specific guardrails and could not be combined dynamically.
The arrival of Claude Opus changed what Mutiny decided to build. Co-founder and CTO Nikhil Mathew saw an opportunity to re-architect the system as an agent-first experience — multimodal, with tool-based access to research, build brand assets, and design the full output. Mutiny built a two-part benchmark to evaluate models: tool-call accuracy, and a creative evaluation using vision capabilities to score outputs across brand alignment, typography, colour, layout, and originality.
After intensive testing with Anthropic’s team, Claude achieved the highest benchmark the company had recorded. Mutiny rebuilt its creation experience around an LLM-native data model and made Claude Opus the default model.
Design satisfaction — measured through in-app user feedback — improved 3x. Asset creation is now 4.5x faster than previous workflows. Nine out of ten sales reps using Mutiny report the product gives them a competitive edge. The agent generates fully branded pitch decks, deal rooms, and business cases from a single website URL, in one shot.
Amazon Bedrock