Make building approachable
A non-engineer should be able to go from idea to a working agent, not a ticket to an engineering team.
Design partner for Augment Group
An AI platform that anyone can build an agent on.
Augment Group is building an AI operating system for institutions, funds, enterprises, sovereigns, and infrastructure operators, where agents act and a human stays in the loop. We designed the platform underneath: how an agent gets created, configured, tested, and trusted.
- Client
- Augment Group
- Partnership
- Ongoing design partner
- Scope
- Brand, agent platform, workflow builder, ops copilots
Everyone wants an AI agent. Almost no one can build one safely.
Institutions are racing to put agents into real workflows: lease review, outage response, diligence, reporting. Most tools force a binary choice, a rigid off-the-shelf chatbot, or a blank canvas that only an engineer can use.
Augment Group’s platform had to hold both ends at once: simple enough for an operations lead to spin up an agent from a template, and precise enough to encode multi-step logic and integrations underneath, all without losing anyone’s trust along the way.
Keep the power underneath
Logic-heavy agents, call routing, transfers, conditional branches, still need node-level control for the people who need it.
Make trust visible
Before an agent touches a real customer or a real lease, someone needs to test it, watch it, and know exactly what it will do.
An assistant that could do a lot, but didn't show it.
Athena Insight, one of the platform’s early agents, opened to a bare prompt and a handful of buttons. Powerful underneath, but nothing on screen told a new user what to ask or what the agent could actually do.
One system, from creating an agent to running one in production.
The same visual language and interaction patterns run underneath the builder, the workflow canvas, and every vertical copilot built on top of it.
We started with the moment that matters most, creating an agent, and worked outward: how it gets a voice, how its logic gets built, where it lives once it’s live, and how the platform proves it’s working before anyone trusts it with something real.
The same components carry the weight everywhere else on the platform: a chat surface for infrastructure operators reading fault data, a lease copilot summarizing risk, a document analyzer pulling structured fields out of a folder of site photos and PDFs.
- Brand identity and logo suite
- Agent builder UX
- Workflow canvas (node-based logic)
- Persona and prompt editor
- Live and simulated testing
- Analytics dashboards
- Vertical copilots (Insight, LeaseIQ, Analyzer)
- Pitch deck design
A guided flow, a prompt, or a full workflow canvas.
An agent starts as a guided flow, a natural-language prompt, or a node-based workflow, plus seven ready-made templates, sales outreach, support, HR onboarding, IT helpdesk, and more, so most teams never start from a blank page.
Plain-language persona, tested in the same breath.
Purpose, character, tone, and guardrails are all plain-language fields, with a live test panel next to the editor, so a change is one click from being heard, not just read.
Branching logic as a diagram, not a settings page.
For agents that need branching logic, call transfers, or system actions, the same builder opens into a node canvas: conversation, function, logic split, SMS, every step laid out where it can be traced and edited directly.
A growing agent roster that still reads as one system.
Every agent an organization runs, voice, chat, or workflow, lands in one workspace: live status, weekly trend, skill count, and a single Create button, so scale never turns into clutter.
The same system, carried into every vertical tool.
The design system carries straight into the copilots Augment Group ships to end users: one entry point across Insight, Hub, and Asset Sense, and a chat surface that turns raw infrastructure data, fault rates, ticket status, regional spread, into an answer with the numbers attached.
A folder of site documents in, a clean template out.
A document analyzer walks a non-technical user through defining an extraction template, then an AI prompt, so uploading a folder of tower site photos and PDFs turns into structured, exportable fields instead of a manual review.
One brand, from the wordmark to the deck.
We built out the Augment Group brand across the product line: a house mark in pink and purple, a distinct wordmark for each product, Hub, Insight, Vision, AI Suite, a gradient identity for hero art and social, and the pitch deck that introduces the platform to investors and partners.
A design system built to keep growing with the platform.
Augment Group is shipping new agents and copilots on the same builder, editor, and dashboard patterns we designed, rather than a new interface for every product line.
This reflects what’s shipped to the platform so far, not a claim about everything on Augment Group’s roadmap.
- 3
- ways to start an agent: guided flow, prompt, or workflow canvas
- 7
- ready-made templates so most teams never start from a blank page
- 4+
- purpose-built copilots shipped on the same design system
- 1
- consistent system from the builder to every vertical tool on top of it
Product names and screens reflect Augment Group’s platform at the time of this writing. Visit Augment Group.
More work
Explore other stories.
Building the interface for an AI platform?
Let's make agents people can actually trust.
Bring the model and the workflows. We'll help you design the interface that makes it usable, and safe, for the people who have to rely on it.