One workspace for every AI model
Synect connects nine AI providers — OpenAI, Anthropic, Google, Perplexity, Cohere, Groq, ElevenLabs, Runway and HeyGen — into a single chat, agent and canvas surface. Pick a model, build a multi-step Spark, edit alongside it in Canvas, share the result.
No credit card. Free during preview. Synect is built and operated in Ireland · GDPR-aligned.
Here's a UK shopping list with current stock at The Pi Hut and Pimoroni:
- ESP32-S3 DevKit-C — thepihut.com/...
- BME280 temp/humidity/pressure — pimoroni.com/...
- 0.96″ I²C OLED — thepihut.com/...
Powered by nine first-party AI providers — no white-label proxies
Three surfaces, one workspace
Synect isn't a chat app pretending to be a productivity tool, or an agent platform pretending to be a chat. It's designed deliberately as three connected surfaces.
Chat
Multi-turn chat with whichever model fits the question. Auto Mode picks the right model when you don't want to. Web search, file attachments, voice input, queued messages, regenerate, branch, share — every behaviour you've come to expect.
Sparks
Build agents with a system prompt, knowledge base, and an optional multi-step pipeline. Sparks can call native web search, fetch pages, run code, generate images, drive Google Slides, or hit any of 80+ Composio integrations (Slack, Notion, Hubspot, Linear, Jira, Salesforce, Stripe…).
Canvas
When the model writes a long document, slide deck, code file, or HTML page, it streams into Canvas — a side-by-side editor where you can revise the model's draft directly. Selection-level edits, full-document rewrites, version history.
Auto Mode
The right model for the question — every time
A two-stage classifier (fast regex rules, then a small LLM fallback) routes every message to the model that handles it best. Web-search-needing prompts go to Perplexity; reasoning questions go to OpenAI o4-mini; coding goes to a fast Gemini; long PDFs go to Gemini 2.5 Pro. You don't pick a model — unless you want to, then explicit override always wins.
Everything you actually use, day one
No feature here is on a roadmap. If it's on this page, it's shipped and you can use it the moment your access is granted.
Live web search
Native web search across OpenAI, Anthropic, Google, Perplexity. With or without the toggle, the model decides — and it cites real URLs.
Long-form artifacts
Markdown, DOCX, PDF, HTML, CSV, code files, PPTX, and Google Slides — all streaming live into Canvas as the model writes.
Media generation
GPT-Image-1 for images, Runway for video, ElevenLabs Music for music, ElevenLabs TTS for narration, HeyGen for spokesperson video.
Realtime voice
Speak with the model in real time — same conversation thread, same context, same Sparks.
Deep research
Long-running multi-source research powered by Perplexity Sonar Deep Research, with citation harvest and stream-back of intermediate findings.
Sparks marketplace
Browse and run pre-built agents — Hotel Persona Generator, SEO competitive review, RFP draft, STL forge — or build your own.
Org & teams
Organisations, teams, role-based permissions, allowed-providers and quality-tier policies, per-org spending controls.
Share & embed
Share any conversation with a public link. Share Sparks across an org. Embed conversations in your own page.
Cross-conversation context
Pull a snippet, a label, or a whole project into a new conversation as context — up to 100k chars, intelligently budgeted.
Auto-titles & labels
Conversations get a 3–5 word title automatically. Labels organise long-running work without you having to.
Ready to stop hopping between apps?
Synect is invite-only while we shape the product with the first wave of teams. Tell us a bit about how you'd use it and we'll get you in.