Akira - Agentic (AI) Personal Finance iOS App (Case Study)
Antony Mumo
Like this project
Posted Sep 18, 2025
Our brief was to automate useful actions with transparent AI, make goals motivating, and connect finances into the tools people already use.
Likes
2
Views
98
Timeline
Jul 1, 2025 - Sep 10, 2025
Clients
Akira
Akira - Agentic (AI) Personal Finance iOS App (Case Study)
Scope
UX & visual system for an AI-first personal finance app that helps users set goals, optimise budgets, and build workflows across everyday tools (Notion, Sheets, Calendar, Slack).
Focus of Case Study
Key design decisions and the process that turned agent behaviour into a simple, motivating experience.
Problem Statement
People want control of money without manual work or judgmental interfaces. Our brief was to automate useful actions with transparent AI, make goals motivating, and connect finances into the tools people already use.
Users and Success Metrics
Primary user
Akira is designed for a 28–38, salaried or freelance professional, plans big and small expenses, uses digital tools to stay organized and avoid surprises.
Key Metrics
Time-to-first goal and budget setting
AI optimisation rate (how frequent a user approves an AI optimisation)
Workflow activation (integration per engaged user)
Budget Optimisation Metrics
Percentage % of AI suggested savings applied
Average Smart Savings by AI per user
Suggestion visibility and removal rate
Design Goals & Principles
Make AI a necessity and not an addition
Be motivating and not punitive
One primary action per surface; details in expanded view
Scannable layouts with clear numerical hierarchy
Integrate with tools users already use
Core Design Decisions
Goals copy
Goal card: Simple layout that shows goal title, progress and badges earned.
AI Insights: Single motivating sentence that sets the users expectations of achieving their goal
Goals Copy & Details
Goal Details
Total saved vs target
Overall progress percentage %
Projected completion date (AI insight)
Goal milestones & badges
Actionable AI suggestions (quick AI fixes)
Budget Optimisation (Smart Savings)
Decision
Continuously surface small, safe reallocations (round-ups, category trims, one-offs) based on spending/saving signals (recurring outflows, deposit patterns, upcoming bills, goal priority).
UI
Compact agent cards with recommendation, one-line “why”, estimated impact, and single CTA (Apply / Alternatives).
Smart Savings
AI Agent Suggestions
AI suggestions are visible by default to accelerate value discovery. Users can remove or disable suggestions on expansion.
Many users benefit from proactive nudges; discoverable opt-out keeps control in the user’s hands. All applied changes are logged and undoable.
Agent suggestions are modular: recommendation + short reason + 1–2 CTAs.
Filters & Discovery
Two-row chips: manual filters (Urgency, Progress, Category, Status) + AI chips (At Risk, Quick Wins, Likely to Achieve, Savings Finder).
AI chips show short tooltips and a “Why?” with top signals.
AI Chip Filters & Suggestions
Workflows & Agents
A workflow is an ordered chain of agents and steps that accomplishes a user goal. It defines triggers, data flow, conditions, and outcomes (e.g., “When paycheck arrives → update goals → push summary to Slack → schedule review in Calendar”).
Agents are single-purpose AI helpers that connect to tools like Plaid, Sheets, Notion, or Calendar to act on financial data.
Essential Agents
Plaid agent - fetches transactions/balances (trigger source)
AI insights agent - summarizes, scores risk/momentum, finds savings
Sheets agent - appends/updates live ledgers
Notion agent - logs journal entries or database rows
Calendar agent - creates review/bill events
Slack agent - pushes digests/alerts to a private channel
Mail/Drive agent - stores exports / sends reports
Automation (Zapier) agents - extends to niche apps
Typical Workflow Patterns
Log + Journal: Plaid → Notion (transaction rows + AI notes).
Digest + Nudge: Plaid → AI Insight → Slack (daily summary + anomaly).
Export + Analyze: Plaid → Sheets (live ledger) → Looker/Drive.
Calendar Review: Plaid → AI Insight (large txn) → Calendar (create review event).
Auto-save: Plaid events trigger Savings Agent round-ups / transfers.
Workflows and Agents
Creating Workflow
Onboarding Flow
The onboarding process is designed to quickly build trust and excitement by focusing on clarity, personalization, and motivation. Instead of overwhelming users with setup tasks, the flow introduces the app’s core value in 3–4 screens.
Onboarding Intro
Account Setup
Loading State & Set Budget and Goals
Closing Note
Designing an app where agents act on users’ finances requires balancing automation with explainability. The core win here is not just having AI find savings, it’s presenting those savings in a way that feels natural, trustworthy, and easy to act on. The design choices above make the product feel less like a finance app and more like a practical, helpful friend that keeps users’ goals on track.
Roles & Responsibilities
Product Strategy & Design - Antony Mumo
Visual Brand Design - Antony Mumo