Mira
A writing partner that knows the brief.
Overall UXE Score
7.96
15 jurors · −0.3 vs season average
- Recognition
- Featured
- Recognized
- 29 Apr 2026
- Season
- Quarter 2, 2026
- Studio
- Softlight, San Francisco
- Categories
- AI & Emerging Experience · Product Experience · SaaS
- Platforms
- Web, Desktop
01/Overview
An AI writing assistant that keeps suggestions inline and reversible, so authors stay in control of voice and intent.
Mira lives inside the document instead of beside it. Select a passage and ask for a rewrite; the suggestion appears in context, with a diff, and nothing changes until the author accepts it.
Every suggestion explains what it did — “more direct, −14 words, reading level grade 9” — so writers learn from the tool rather than outsourcing to it.
Documents can carry a brief (audience, tone, key messages) that Mira uses for every suggestion, so a team’s voice stays consistent.
- Client
- In-house product
- Audience
- Marketing, product and communications teams
- Industry
- AI Products
- Platforms
- Web, Desktop
- Status
- Public beta since January 2026
- Markets
- Worldwide
- Languages
- 3 — English, Spanish, German
02/Honors
Recognized by the editors. Awarded by the jury.
Recognition
Featured
Meets the UXE standard. Earns a public profile, a badge and a place in the quarterly awards.
Conferred by the editors on 29 Apr 2026 · minimum UXE 7.0
Awards · Quarter 2, 2026
Season results- AI & Emerging ExperienceHonorable Mention
11 / 12
- Product ExperienceNominee
01 / 12 · SaaS
Results announced 15 Jul 2026 by the category juries.
03/Jury verdict
In the jury’s words.
“Mira keeps the author in charge. Small, explained, reversible suggestions are how AI writing should work.”
Chloé Martin
Content Design Lead, Plume & Co · Canada
On behalf of 15 jurors
04/Score breakdown
How the jury scored it.
- Overall UXE Score−0.3 vs Q2 2026 average
- 7.96
- Strongest criterionUI & Consistency
- 8.3
- Jury agreementaverage spread ±0.38
- Moderate
- Jurorsof Q2 2026’s panel
- 15
- 8.2−0.3
Usability
Can people complete what they came to do — efficiently, confidently and without error?
- 7.9−0.3
Information Architecture
Is content organized, labeled and navigable in the way people actually think?
- 8.3−0.3
UI & Consistency
Is the visual language coherent and purposeful across every screen and state?
Jury noteCalm, coherent interface that lets the document stay the hero.
- 8.0−0.4
Interaction & Feedback
Does every action respond clearly, promptly and meaningfully?
- 7.8−0.5
Performance
Does it feel fast — perceived and measured — on real devices and networks?
Jury noteSuggestions can take close to three seconds — long enough to break flow.
- 7.4−0.4
Accessibility
Can everyone use it, including people relying on assistive technology?
Jury noteThe floating toolbar obscures text at high zoom levels.
- 8.1−0.2
Learnability
How quickly does a first-time user become a confident one?
Scores are the mean of 15 jurors. Ticks mark the Q2 2026 average across 4 recognized projects.
05/The jury
15 ballotsEvery vote, on the record.
Each juror scored Mira independently, from 1 to 10 on every criterion. Nothing is weighted by seniority and no score is discarded — the published result is simply their mean.
One juror Jury mean
| Juror | Usab | IA | UI | Inter | Perf | A11y | Learn | Overall |
|---|---|---|---|---|---|---|---|---|
| Ana CostaInclusive Design Researcher, Maré Digital · PT | 9.0 | 8.4 | 8.6 | 8.4 | 7.8 | 7.7 | 8.1 | 8.29 |
| Min-jun ChoDesign Lead, Seoul Form · KR | 8.2 | 7.9 | 8.6 | 8.7 | 8.7 | 7.5 | 8.4 | 8.29 |
| Amara NwosuDesign Director, Kestrel & Vale · GB | 8.2 | 8.2 | 8.7 | 8.3 | 7.8 | 7.4 | 8.6 | 8.17 |
| Priya RamanHead of Product Design, Tessera Cloud · IN | 8.4 | 7.7 | 8.3 | 8.2 | 7.6 | 8.0 | 8.7 | 8.13 |
| Olivia GrantVP of Design, Harbor & Pine · US | 8.3 | 8.1 | 7.9 | 9.0 | 7.6 | 8.0 | 7.9 | 8.11 |
| Chloé MartinVerdictContent Design Lead, Plume & Co · CA | 7.6 | 7.9 | 8.4 | 8.1 | 8.1 | 7.9 | 8.4 | 8.06 |
| Ravi MenonWeb Performance Engineer, Quickline · SG | 8.1 | 7.8 | 8.2 | 7.9 | 8.5 | 7.5 | 8.1 | 8.01 |
| Lukas BrennerPrincipal Product Designer, Feldwerk · DE | 8.9 | 8.5 | 8.1 | 7.2 | 8.5 | 7.1 | 7.6 | 7.99 |
| Nina PetrovaDesign Director, Graphit · BG | 8.2 | 7.9 | 8.5 | 7.4 | 7.9 | 7.3 | 7.9 | 7.87 |
| Camille RouxCreative Director, Atelier Nuit · FR | 7.9 | 7.9 | 8.6 | 7.6 | 7.2 | 7.2 | 8.6 | 7.86 |
| Liam O’ConnorAccessibility Engineer, Cardinal Web · IE | 8.5 | 7.9 | 8.2 | 7.9 | 7.6 | 7.4 | 7.4 | 7.84 |
| Mei TanakaUX Research Lead, Hoshi Labs · JP | 7.9 | 7.8 | 8.2 | 8.0 | 7.8 | 7.1 | 8.0 | 7.83 |
| Kwabena AsanteService Design Lead, Adinkra Digital · GH | 7.8 | 7.5 | 7.9 | 8.3 | 7.6 | 7.2 | 8.0 | 7.76 |
| Sofía MarínAccessibility Lead, Clarosol · ES | 8.2 | 7.8 | 8.4 | 7.4 | 6.5 | 7.1 | 7.8 | 7.60 |
| Marco BelliniSenior Product Designer, Studio Vetro · IT | 7.8 | 7.2 | 7.9 | 7.6 | 7.8 | 6.6 | 8.0 | 7.56 |
| Jury mean | 8.2 | 7.9 | 8.3 | 8.0 | 7.8 | 7.4 | 8.1 | 7.96 |
| Spread (σ) | ±0.38 | ±0.31 | ±0.26 | ±0.49 | ±0.52 | ±0.37 | ±0.36 | ±0.22 |
marks each juror’s area of expertise. Deeper tint, higher score.
01
Submitted
6 Apr
Entry received, free of charge
02
Jury scoring
12 Apr – 28 Apr
15 independent ballots
03
Recognized
29 Apr 2026
Featured
04
Awards
15 Jul 2026
Honorable Mention · AI & Emerging Experience
06/Screens
1 screenThe product, screen by screen.
07/Case study
From problem to proof.
The challenge
AI writing tools rewrote whole documents at once, flattening voice and leaving authors unsure what had changed.
The approach
Softlight made every AI action small, local and reversible: selection in, suggestion with a diff out, and a reason attached.
Research
Writer interviews
30 professional writers and editors
Comparative testing
Mira vs. two chat-based tools, 24 participants
Beta telemetry
4,500 documents, opt-in
Key insights
- 01
Writers accept help paragraph by paragraph, not document by document.
- 02
An explanation turns a suggestion into a lesson.
- 03
A shared brief matters more to teams than a smarter model.
Outcomes
68%
Suggestions accepted
vs. 31% for whole-document rewrites
+2.1
Voice consistency
editor-rated, 1–10 scale
−35%
Editing time
first drafts to approved copy
Outcomes are reported by the submitting team and reviewed by the jury.
08/Accessibility & performance
Built for everyone, on any device.
Accessibility
WCAG 2.1 AA
Audited by In-house audit, January 2026
7.4
Jury
- Suggestions are announced with their diff to screen readers
- All AI actions reachable from the keyboard
- Reading-level feedback on every suggestion
Known gaps
- Floating toolbar can obscure text at 200% zoom
- Gradient backgrounds reduce contrast in the marketing site
Performance
3 of 4 within budget
MacBook Air M1, cable connection, lab median
7.8
Jury
| Metric | Measured | Budget |
|---|---|---|
| Largest Contentful Paint | 1.9 s (within budget) | ≤ 2.5 s |
| Interaction to Next Paint | 180 ms (within budget) | ≤ 200 ms |
| Cumulative Layout Shift | 0.03 (within budget) | ≤ 0.1 |
| Suggestion latency | 2.8 s (over budget) | ≤ 2.0 s |
Within budget Over budget
09/Design system
Quill, the system behind Mira.
- System
- Quill
- Components
- 40
- Design tokens
- 110
Color
- #18181B
Graphite
- #FB923C
Apricot
- #F472B6
Rose
- #A78BFA
Lavender
- #E9E5FF
Highlight
Typography
Inter
Interface
Google Fonts
Newsreader
Document text
Google Fonts
UX patterns
- Inline AI
- Diffs
- Undo
- Briefs
- Collaboration
10/Credits
The people who made it.
Studio
Softlight
San Francisco, United States
An AI product company focused on writing tools that keep the author in charge.
- Founded
- 2021
- Team
- 11–50
- Recognized
- 1
Team
- Rachel StoneHead of Design
- Kenji WatanabeProduct Designer
- Maya PatelContent Designer
- Chris AlvarezEngineering Lead
- Client
- In-house product
- Industry
- —
Built with
- React
- ProseMirror
- TypeScript
- Claude API
- Electron
- Figma
→/More from the showcase
Whole season
