Memo
AI
ModelsHow it worksJourneyBuilt byContact
Sign in
17 months · from prototype to v2.0145

The journey.

From a rough side-project to a production AI assistant used daily at Memo Fashion. 17 months of research, engineering and refinement — all done in-house.

November
2024

Research begins

Early experiments with open-source language models. Comparing architectures, evaluating output quality, measuring latency.

R&Dv0.1
December
2024

First internal prototype

A bare-bones chat interface wired to a single model. No streaming, no memory — but it could hold a short conversation.

Prototypev0.3
January
2025

Multi-model cascade architecture

Moved from a single model to a fallback chain. If one engine is rate-limited, the next takes over seamlessly.

Resiliencev0.5
February
2025

Streaming responses

Word-by-word output. Tokens appear as they're generated. Server-Sent Events pipeline built from scratch.

Streamingv0.7
March
2025

Session memory & history

Conversations now persist across reloads. Multiple chats per user, full history, auto-generated titles.

Memoryv0.9
April
2025

Document understanding

PDF reading capability added. Upload an invoice, a contract — AI extracts and summarises.

PDFv1.0
May
2025

Vision capability

Image understanding added. Describe a product photo, OCR a receipt, analyse a moodboard.

Visionv1.2
June
2025

Web search integration

Live web search triggers automatically on current-events questions. Real-time knowledge.

Live datav1.3
July
2025

Image generation engine

Generate visuals from text prompts. Built to handle unlimited requests at zero cost.

Image genv1.4
August
2025

Excel & Word I/O

Ask for a table — get a real downloadable .xlsx file. Ask for a letter — get a .docx.

File I/Ov1.5
September
2025

Voice input

Speak your message — local transcription. Fast, private, accurate.

Voicev1.6
October
2025

Custom instructions & tone

Personal context applied to every chat. Tone selector for every reply.

Personalv1.7
November
2025

PWA & mobile experience

Installable on phones, tablets, desktop. Offline-capable. iOS notch handling.

PWAv1.8
December
2025

Multi-file attachments

Attach up to 10 files per message. Mix images, PDFs, Excel, Word.

Multi-filev1.9
January
2026

PDF export pipeline

AI-generated content downloadable as real A4 PDFs. Print-ready, branded.

PDF exportv1.95
February
2026

Temporary chat mode

Private chats that don't save anywhere. For when you need to think out loud.

Privacyv1.98
March
2026

Closed beta with the Memo team

Internal rollout to early testers. Feedback gathered, bugs squashed, UX refined.

Betav2.0
13 Apr
2026

Company-wide trial launch

Memo AI rolled out to every member of the Memo Fashion team on 13 April 2026.

Trialv2.01
Today
2026

Dedicated home at ai.memo.co.uk · v2.0145

Memo AI now lives on its own domain with a premium interface. Current production version: 2.0145.

Live nowv2.0145
17
Months in development
19
Major releases
2.0145
Live version today
© 2026 Memo Fashion Limited · Powered by Sarma Linux
PrivacyTermsContact