Every strong design decision traces back to two things: what your users actually said, and what your product actually does. The problem isn't that teams lack this information — it's that it lives in two places that rarely talk to each other. Meeting notes sit in Granola. Behavioral signals sit in your analytics. And somewhere in between, a designer is expected to stitch it all together from memory.
Septra's Product Designer agent closes that gap. It reads your real research and your real data, then produces design artifacts you can hand straight into your design tool — so your Figma frames are grounded in evidence, not instinct.
The gap between research and design
Ask any product designer how much of their research actually makes it into the final screen and you'll get an honest, slightly uncomfortable answer: *some of it, if they remember.*
User interviews get transcribed but never re-read. Analytics dashboards get checked during discovery and forgotten by the time wireframes ship. By the time a screen reaches Figma, it's been filtered through days of context-switching and a growing pile of assumptions.
This is the research-to-design gap, and it's expensive in both directions: designs that ignore what users said, and designs that ignore what the data showed.
How the Product Designer agent works
Septra deploys a Product Designer agent — we call her Lyra — that sits on top of your workspace's private knowledge base. That knowledge base isn't a generic corpus; it's built from the tools your team already uses. And Lyra runs on a higher-reasoning model, because design synthesis is exactly the kind of work that rewards deeper reasoning over raw speed.
The critical difference from a generic AI design prompt: Lyra doesn't guess. Every artifact she produces is grounded in two connected sources.
1. Meeting intelligence from Granola
Lyra has real-time access to your Granola meeting data — transcripts, notes, and summaries. When a customer says "I can't find the export button" or a stakeholder says "onboarding feels too heavy," that signal is no longer trapped in a note nobody reopens. Lyra queries it directly, pulls the exact language users used, and carries it into the design.
2. Product analytics from your stack
The other half is behavioral truth. Septra syncs your analytics — Google Analytics, Google Ads, Meta Ads, LinkedIn Ads, email, and social — into the same knowledge base through its intelligence pipelines. Lyra can see where users actually drop off, which features get adopted and which get ignored, and where the funnel leaks.
What people say in interviews and what they actually do in the product are often two different things. Lyra sees both, side by side.
The artifacts Lyra produces
This is where the workflow stops being "chat" and starts being design work. Lyra uses an industry-standard design-artifact glossary to turn raw research into structured deliverables you can actually use:
- Personas — synthesized from real interview language and behavioral segments, not invented stereotypes. Each persona cites the meetings and data points it came from.
- Journey maps — the full user path, with the friction points your analytics already flagged, not the ones a generic AI assumed.
- Wireframe specs — screen-by-screen structure with layout, hierarchy, and states, ready to translate into frames.
- Typography and visual style direction — grounded in your brand's established identity, not a fresh guess every time.
The point isn't that Lyra replaces your designer. It's that she compresses the research phase from days to minutes and hands over artifacts that are already evidence-backed — so the designer's time goes into craft, not re-reading transcripts.
From artifact to Figma, via the Septra MCP server
Artifacts are only useful if they reach the tool where the actual design happens. That's where the Septra MCP server comes in.
Septra exposes a Model Context Protocol (MCP) endpoint — a standards-based interface that lets external tools and AI assistants talk to your workspace securely. Through it, your design environment can pull the same artifacts, knowledge-base documents, and intelligence sources Lyra used.
Here's the practical flow:
- Lyra ingests Granola notes and analytics, and produces a set of design artifacts.
- Those artifacts are stored in Septra's schema-agnostic artifacts system — versioned and reviewable, not lost in a chat log.
- From Figma, or an MCP-capable assistant connected to your design workflow, you query the Septra MCP server for the persona, journey map, or wireframe spec you need.
- The evidence — the exact user quote, the exact drop-off metric — arrives in your design tool, ready to inform the next frame.
The result: your Figma canvas is no longer a blank artboard plus a vague prompt. It's a blank artboard plus the actual research. Every design decision can trace back to something a user said or something the data showed.
Why this produces better designs
Design quality has never been about how many pixels you can produce. It's about whether the pixels reflect reality.
Grounding design in research doesn't make it slower — it makes the iteration loop honest. When a persona says "this user is price-sensitive," you know it's backed by a sales call in Granola, not a cliché. When a wireframe removes a step from onboarding, you know it's backed by a funnel drop-off in your analytics, not a hunch.
That's the real promise of agentic design: not that an AI does the design for you, but that the design it helps you make is *true*.
Get started
Connect Granola and your analytics to Septra, deploy the Product Designer agent, and ask Lyra for a persona or a journey map for the product area you're working on. Then connect your design workflow through the Septra MCP server and pull those artifacts straight into Figma. You'll spend less time reconstructing research — and more time designing from it.
Frequently Asked Questions
Does Lyra replace my human product designer?
No. Lyra compresses the research and synthesis phase — producing evidence-backed personas, journey maps, and wireframe specs — but the final craft, visual polish, and judgment still belong to your designer. Lyra hands them a head start, not a finished screen.
What data does the Product Designer agent actually use?
Two sources, both connected to your Septra workspace: Granola meeting transcripts, notes, and summaries on the qualitative side, and your product analytics — Google Analytics, Google Ads, Meta Ads, LinkedIn Ads, email, and social — synced into your knowledge base on the behavioral side.
How do the artifacts get into Figma?
Through the Septra MCP server. It exposes your workspace's artifacts and knowledge base over the Model Context Protocol, so any MCP-capable design environment or assistant can pull the persona, journey map, or wireframe spec directly into your workflow.
Do I need to be technical to set this up?
No. Connecting Granola and your analytics, then deploying the Product Designer agent, happens from the Septra interface. The MCP endpoint handles the rest — no custom integration code required.
