Powered by Google's Most Advanced AI
Aida orchestrates a suite of Google's frontier AI models through Firebase Genkit — each chosen for a specific role in your marketing workflow. Here's exactly which models run under the hood and what they do.
Orchestrated with Firebase Genkit
All AI model calls in Aida are managed through Firebase Genkit — Google's open-source AI orchestration framework. Genkit provides structured input/output validation via Zod schemas, flow tracing, and unified model access across Vertex AI and Google AI. Every flow in Aida is defined, typed, and observable through Genkit.
Gemini
via Google Cloud Vertex AI & Google AI
Gemini is Google's most capable family of multimodal AI models. Aida uses several Gemini variants across its platform, choosing the right model for each task.
How Aida uses it
- Brand Brain analysis — deep reasoning over your website, tone, and market position
- Campaign strategy generation — Gemini 2.5 Pro produces full multi-phase campaign plans
- Post and caption writing — Gemini Flash generates platform-optimised social content at speed
- Content DNA extraction — Gemini identifies tone, hook format, topic, and emotional register from approved posts
- Performance analysis — Gemini reads engagement data and writes actionable playbooks
- Brand interview — conversational intelligence that builds your Brand Brain interactively
Models
gemini-2.5-progemini-2.5-flashgemini-3.1-progemini-3-flash-previewgemini-2.5-flash-liteImagen
Image Generation via Google AI
Imagen is Google's state-of-the-art text-to-image diffusion model. Aida uses Imagen to generate on-brand visuals for every piece of social content — no design skills required.
How Aida uses it
- Campaign image generation — creates images for Instagram, Facebook, LinkedIn, and TikTok posts in the correct aspect ratio (1:1, 4:5, 9:16, 16:9)
- Moodboard-driven visuals — colour palette, lighting, and aesthetic from your brand moodboard are passed directly to Imagen as structured style parameters
- Google Ads PMax asset images — landscape images for Performance Max asset groups
- Carousel slide images — one image per slide, generated in parallel
- Brand safety layer — every Imagen output is screened by Cloud Vision SafeSearch before delivery
Models
imagen-4.0-generate-001imagen-4.0-ultra-generate-001imagen-3.0-generate-002Chirp
Speech & Audio via Gemini TTS
Chirp is Google's family of high-quality neural voices, available through Gemini's text-to-speech capability. Aida uses Chirp voices to turn AI-written scripts into professional-grade podcast audio.
How Aida uses it
- Podcast generation — Aida's Podcast Creator writes a multi-speaker script with Gemini, then uses Chirp voices via Gemini TTS to synthesise each speaker turn into natural-sounding audio
- Multi-speaker conversations — different Chirp voices are assigned to host and guest roles for realistic dialogue
- Brand audio content — businesses can publish AI-generated podcast episodes directly to their marketing channels
Models
gemini-2.5-flash-preview-tts (Chirp HD voices)Lyra
Neural Audio Codec via Google AI
Lyra is Google's neural audio codec — a low-bitrate, high-fidelity audio compression model. The audio generated by Aida's podcast and TTS features is encoded and delivered using Lyra-based compression, ensuring high-quality audio even at reduced file sizes.
How Aida uses it
- Audio encoding for podcast output — podcast episodes generated by Aida's TTS pipeline are compressed and streamed using Lyra-compatible encoding
- Efficient audio delivery — Lyra enables Aida to deliver podcast audio files that are both high quality and appropriately sized for web delivery and social sharing
Models
Lyra v2 (via Gemini TTS audio pipeline)Veo
Video Generation via Google Cloud Vertex AI
Veo is Google's advanced video generation model. Aida uses Veo to transform brand images and text prompts into short-form video clips — ready for Reels, TikTok, and Stories.
How Aida uses it
- AI video clip generation — turn any image or text description into a polished short video
- Animated brand content — bring still images to life for social media
- Veo 3 audio-enabled video — generate videos with native AI-synthesised audio tracks
Models
veo-2.0-generate-001veo-3.0-generate-001Cloud Vision
Image Safety via Google Cloud Vision AI
Google Cloud Vision provides powerful image analysis capabilities. Aida uses Vision's SafeSearch Detection on every AI-generated image before it is shown to users.
How Aida uses it
- Brand safety screening — every image generated by Imagen is checked against Vision SafeSearch before delivery. Images flagged as adult, violent, or racy are silently regenerated with additional safety constraints
- Fail-open design — if the Vision API is unavailable, Aida fails open to never block content delivery, while logging the skipped check for review
Models
Vision API SafeSearch DetectionAida is a Google for Startups portfolio company