Definition launches ‘Try before you AI’

The launch of the demo suite comes at a critical time. Harvard Business School research shows that AI makes teams more productive and improves the quality of their work. Consultants using AI complete 12.2% more tasks on average, 25.1% more quickly and to a 40% higher standard. Yet data confidentiality concerns mean that more than a quarter (27%) of organisations have temporarily banned public gen AI applications – and so aren’t seeing those benefits – according to Cisco’s latest Data Privacy Benchmark report.

The ‘Try before you AI‘ suite addresses privacy concerns and indecision over model selection that have hindered AI implementation. It offers access to the world’s most powerful language, image, audio, and vision AI models from OpenAI, Google, Anthropic, and AWS, all in one secure place.

Luke Budka, AI director at Definition, notes: “Businesses need to make sure their AI investments deliver. We believe this is achieved by really understanding which AI is best for different activities and then equipping clients with bespoke prompts and fine-tuned models, designed to address their specific business use cases. In future this approach will extend to autonomous AI that can carry out activities independently.”

Louise Vaughan, CCO and co-founder at Definition, adds: “Using the potential of AI – both in client service and in products that help brand-side marketing teams – is a key part of our strategic plan. Companies without internal AI tools are already falling behind, but their concerns over privacy and challenges over which models to commit to are real. This personalised AI suite – available to trial – solves those problems.”

Definition’s demo suite includes GPT-4o, Claude 3.5 Sonnet, AWS Transcribe, Whisper, DALL-E 3, and Imagen 3.

The suite also features a built-in prompt library which can be tailored to each organisation, with prompts written by Definition’s language and design experts. And OpenAI’s File Search and Code Interpreter tools, enabling clients to augment the AI models with their own data and analyse spreadsheets, PDFs, and Word documents efficiently. The fully private, encrypted environment ensures all entered and uploaded data remains confidential and isn’t used for training purposes.