Amanat Studio designs the trust layer for AI agents in fintech, identity, insurance and healthcare: the approvals, limits, undo and audit trails that turn a clever demo into a product people rely on.
TRANSPARENCY · REVERSIBILITY · USER CONTROL · SUPERVISION · TRACEABILITY
01 — THE TRUST GAP
That gap is where trust dies. Users can’t see, can’t stop, and can’t undo, so they switch the agent off.
WHAT YOUR AGENT DID
09:02:11
read
accounts.balance → $2,410.08
09:02:13
fetch
bills.pending → 5 items
09:02:14
match
payee “City Electric” → new ID
09:02:15
calc
amount $142.60 (+38% vs avg)
09:02:16
decide
within policy? assumed yes
09:02:17
execute
payment.create($142.60)
09:02:19
execute
payment.create($59.99)
09:02:20
notify
“Task completed”
+ 6 more steps…
WHAT YOUR USER SAW
✓
Bill Agent · now
Task completed
02 — THE TRUST FRAMEWORK
Five pillars. One scale of risk. Every screen we design is checked against both.
Transparency
Shows its plan, progress and reasons in plain language.
“What is it doing, and why?”
Reversibility
Drafts before sending. Undo windows after acting.
“Can I fix a mistake?”
User control
Limits, permissions and autonomy levels set by the user.
“Am I still in charge?”
Supervision
High-stakes actions wait for approval, with full context.
“Does it ask first?”
Traceability
A readable log of every action, reason and approver.
“What happened, exactly?”
THE RISK LADDER
The higher the stakes, the stronger every pillar must be
L1
Observe
Act freely, log quietly
L2
Suggest
Prepare; user reviews
L3
Act · reversible
Act, notify, offer undo
L4
Act · high stakes
Always ask first
L5
Critical
Approve, confirm, human fallback
03 — SELECTED WORK
Concept studies · 2026
L4 · APPROVAL NEEDED
City Electric Co.
$142.60
vs last month
+38%
Approve
Edit
FINTECH
CONCEPT
A bill-pay agent people trust with their money
THE PROBLEM
Payments ran silently. Users saw “Task completed” and turned the agent off after one surprise charge.
WHAT WE DESIGNED
User-set limits, approval cards with context and comparisons, a 30-minute undo window, and a plain-language activity log.
Supervision
Reversibility
Traceability
Read case study →
READY TO SUBMIT · 1 TO CHECK
Auto claim · Rear collision
Photos
4 attached ✓
Police report
Matched ✓
Repair estimate
$2,340 · estimated
Review estimate
Ask a person
INSURANCE
CONCEPT
A claims agent that files for you, and shows its work
THE PROBLEM
The agent filled claims automatically, but users couldn’t tell which details it had guessed. Wrong claims meant long disputes.
WHAT WE DESIGNED
A review screen that marks every field the agent estimated, source links for each document, and a human handoff for disputed claims.
Transparency
Supervision
User control
Read case study →
L5 · CONSENT REQUIRED
Share records with Riverside Clinic?
Includes
Labs, prescriptions
Excludes
Therapy notes
Access ends
In 30 days
Allow access
Customize
HEALTHCARE
CONCEPT
A records agent that asks before it shares
THE PROBLEM
The agent sent full medical histories to new clinics. Patients never knew what was shared, with whom, or for how long.
WHAT WE DESIGNED
Consent cards listing exactly what is included and excluded, time-limited access, and a revoke button in the sharing log.
User control
Supervision
Traceability
Read case study →
04 — HOW WE WORK
Fixed prices. No surprises. You approve every step.
01
DAY 1 · FREE · 20 MIN
Discover
A short call to understand your agent, your users, and where they hesitate.
YOU GIVE
↑ A quick demo or screen recording
↑ What your agent does, and for whom
YOU GET
↓ An honest read of your biggest trust risk
↓ A clear recommendation: audit, sprint, or neither
02
5–7 DAYS · FIXED PRICE
Audit
We test every agent action against the TRUST Framework and the Risk Ladder.
YOU GIVE
↑ A test account or screenshots of key flows
↑ A list of what the agent can do
↑ An NDA, if you need one
YOU GET
↓ Your TRUST Score across five pillars
↓ Action map and prioritized critical fixes
↓ A 45-minute walkthrough with your team
03
2–3 WEEKS
Design sprint
We redesign the flows that matter most and build the trust components.
YOU GIVE
↑ Two check-ins a week
↑ Feedback within 48 hours
YOU GET
↓ Redesigned approval, undo and activity flows
↓ A reusable trust component set in Figma
↓ A clickable prototype
04
FINAL WEEK
Handoff
Everything your developers need to build it right the first time.
YOU GIVE
↑ An hour with your developers
YOU GET
↓ Specs, states and edge cases
↓ A recorded video walkthrough
↓ Two weeks of questions answered
05
OPTIONAL · MONTHLY
Partner
We stay on as your trust design partner as the agent grows.
YOU GIVE
↑ Your release roadmap
YOU GET
↓ A trust review before every release
↓ New agent features designed
05 — PRICING
Trust Audit
START HERE
from
one-time · 5–7 days
Teams with a live or beta agent who want to know what’s breaking trust.
WHAT YOU GET
✓ TRUST Score across all five pillars
✓ Every agent action mapped to the Risk Ladder
✓ Top 3–5 critical fixes, each with a redesigned screen
✓ Prioritized roadmap: now, next, later
✓ PDF report your whole team can use
✓ 45-minute walkthrough call
Fee credited in full if you book a sprint within 30 days
Agent UX Sprint
from
per sprint · 2–3 weeks
Teams ready to fix the flows that matter most before scaling.
WHAT YOU GET
✓ Everything in the Trust Audit
✓ Redesigned approval, undo, permission and activity flows
✓ Reusable trust component set in Figma
✓ Clickable prototype for user testing
✓ Developer specs, states and edge cases
✓ Two revision rounds + 2 weeks of support
Most teams need one sprint; complex agents may need two
Design Partner
from
per month · cancel anytime
Teams shipping new agent features every month.
WHAT YOU GET
✓ Ongoing design for new agent features
✓ Trust review before every release
✓ Priority async support in Slack
✓ Monthly strategy session
✓ Quarterly TRUST Score re-audit
Minimum commitment: 2 months
PAYMENT
50% to start, 50% on delivery
PAY BY
Bank transfer, Payoneer or Upwork
CONFIDENTIAL
NDA available before we see anything
GUARANTEE
Not useful? We refine it free until it is
Something entrusted to your care, to be protected and returned intact.
Money
Identity
06 — WHY AMANAT
Their money. Their identity. Their health records. Every time someone lets an agent act for them, they trust it with something that matters. We design agents that are worthy of that trust, and that give it back intact.
Why teams choose Amanat
Proven at national scale
Identity services for millions of citizens: low literacy, many languages, zero room for error. Agent trust needs the same discipline.
Trust is all we do
Not a generalist agency. Every project, framework and checklist is built around one question: will people rely on it?
Designs that ship
Design and front-end skills in one team. We know how agents are built, so our work is buildable, not just beautiful.
Senior work, fair pricing
Senior-level design from Islamabad, at a fraction of typical US and European agency rates.
THE PRINCIPLE BEHIND EVERYTHING WE DESIGN
07 — START HERE
↑ WHAT WE’LL ASK FROM YOU
2 minutes to fill in this form
A 20-minute call with a product lead
Access to your agent: a test account, screenshots or a screen recording
↓ WHAT YOU GET FROM US
A reply within one working day
An honest read of your trust risks, even if you don’t hire us
A fixed-price proposal within 48 hours of the call
An NDA before we see anything confidential
WHAT HAPPENS AFTER YOU HIT SEND
1
WITHIN 24 HOURS
We reply with a link to book your call
2
20-MINUTE CALL
We look at your agent together
3
WITHIN 48 HOURS
You get a fixed-price proposal
4
WORK BEGINS
After the NDA and 50% deposit