Image generation
AIMY Avatar
One photograph, a consistent set of portraits.
Staff and event portraits generated from a single photograph in a house style you define — the same lighting, background and framing across an entire organisation, without booking a photographer for every new joiner.
- One photograph per person, taken on a phone
- The same lighting, background and crop across the whole roster
- A person approves every portrait before it is published
What deployments look like
- One photo
- per person — no studio session to schedule
- phone camera is sufficient
- One style
- applied identically across the organisation
- agreed with you once
- Re-runnable
- restyle the whole roster when branding changes
- without a re-shoot
Everyone's photo looks different, because everyone's photo is different.
A staff directory assembled over ten years: some studio shots, some cropped out of a wedding, some taken against a beige wall in 2016. It reads as untidy on a website and worse on a proposal. The usual fix is to book a photographer, get most of the office on the day, and then repeat it for every new joiner.
- A photographer costs a day and still misses whoever was on leave.
- New joiners never match the set, so the set degrades from the day it is made.
- Consumer avatar apps require uploading staff photographs to a service nobody vetted.
- Styles drift when different people edit images at different times.
What AIMY Avatar actually does.
A house style, agreed once
Lighting, background, crop, aspect and treatment are defined with you at the start, then applied identically to everyone.
The model is a configuration decision
Image generation moves quickly, so we select the model that fits your quality bar, your licensing needs and where you are deploying — and we can change it later without you re-agreeing the house style.
Consistent across a roster
The point is not one striking portrait. It is two hundred portraits that look like they were taken on the same afternoon.
Photographs stay where you put them
Deployed in your cloud tenancy or on GB10, so staff photographs need never leave the boundary you already trust.
Consent recorded, provenance kept
Every portrait is tied to its source photograph and to the consent that permitted it. A likeness is not generated for anyone who has not agreed.
Human approval before publication
Nobody's portrait goes out unreviewed. The approval step is part of the workflow, not an optional extra.
From one photograph to a portrait you have approved.
-
01
Agree the style
A short session and a few test portraits until the look is right. This is the only subjective part of the process.
-
02
Collect
One photograph per person, with consent recorded at the same time. A phone camera against any wall is enough.
-
03
Generate
Portraits are produced in the agreed style, at the sizes and crops your directory, badges and decks need.
-
04
Review
Someone approves each portrait — and each person sees their own before it is used anywhere.
-
05
Publish and repeat
Approved portraits go to your directory or site. New joiners run through the same path, so the set never drifts.
Technical profile
- Deployment
- Managed cloud · Customer VPC · On-premise on GB10
- Model
- Selected per engagement and licensed for your commercial use
- Input
- One photograph per person; a phone camera is sufficient
- Output
- PNG or JPEG at the sizes, crops and aspects you specify
- Style
- Defined with you at the start and applied consistently thereafter
- Scoping
- Per roster, delivered on enquiry
Security & data handling
- Source photographs stay inside the boundary you deploy into
- Consent recorded against every person before any portrait is generated
- A person reviews and approves every portrait before it is published
- Each subject sees their own portrait before it is used
- Configurable retention for both source photographs and generated output
- No customer photographs used to train third-party models
The ones procurement always asks.
Something not covered here? Ask directly — you will get a straight answer, including when the answer is that we are not the right fit.
Ask usWhich model does this use, and can we use the output commercially?
Model choice is part of scoping rather than a fixed property of the product — the field moves too quickly to commit to one. What does not vary is the licensing: several of the strongest open-weight image models are released under non-commercial terms, so we make sure whatever we deploy is licensed for your commercial use, and that you have that in writing.
Do we need consent from staff?
Yes, and the process is built around it. A photograph of an identifiable person is personal data, and a portrait made from it is a use of their likeness. Consent is recorded before generation and each person sees their own portrait before it is used.
Can it place someone somewhere they never were?
That is not what this is, and it is not something we build. AIMY Avatar restyles a portrait: same person, consistent presentation. We do not build capability aimed at misrepresenting where someone was or what they did.
Is this part of AIMY Expert?
No. It is a separate application that shares the brand and the deployment options. It does not read your document corpus and does not require AIMY Expert.
Sectors where AIMY Avatar is deployed
Also on the AIMY Platform
Agentic process automation
AIMY Agents
Multi-step work, executed inside boundaries you can defend.
Explore AIMY AgentsDocument processing
AIMY Extract
Paper in, structured records out.
Explore AIMY ExtractEvent access
AIMY Smart Face
Register once, then walk in.
Explore AIMY Smart FaceShow us the directory you are embarrassed by.
A first conversation covers the style you want, how many people are in scope, and how consent will be collected. The licensing question we handle.