Platform
The whole job, from brief to delivery.
Per-client agents, a governed generation loop, and real deliverables — inside the workflow your agency already runs.
How it works
From client context to approved work.
Tempera puts the client’s history, the brief, the work, and the decision in one governed loop. Every review improves what comes next.
Give each client its own context
Point Tempera at the material you designate — brand guidelines, approved and rejected work, correspondence, research, rate cards, and campaign files. Each client’s agent is grounded in its own context — never another client’s.
Produce work inside the brief
Copy, briefs, research, social, graphics, media plans, and AV materials are created as structured deliverables — not disconnected chat responses.
Keep a human accountable
Enforced in code: the moment a second person is active, whoever ran the work can’t approve it, and one accountable owner signs off before anything is client-ready.
The loop
Brief in, approved work out.
One pass, four stations — and nothing moves forward without clearing the one before.
Brief
A chat turns the request into a structured brief, scaffolded per deliverable type.
Generate
The client's agent produces options, grounded in that client's own context and routed to the right model.
Review
A human reviews every option. Add a second active user and whoever ran the work can't approve it.
Deliver
Approved work ships in its native format — Docs, Slides, PDF, PSD, XLSX — into your own storage.
Every verdict feeds the next generation
Built for agencies
The hierarchy your agency already runs on.
Work stays under the client, brand, campaign, and activity — with the owner and review path visible. Tempera fits the operating model instead of replacing it with a chat window.
One accountable owner
Every client, campaign, and piece of work has exactly one name against it. Reviews route to that person.
Effort scales with the job
A quick adaptation stays light; an integrated campaign convenes the full cast. The platform never forces ceremony onto small jobs.
Real deliverables
Not chat responses. Deliverables.
Structured outputs with the integrity checks your clients' money deserves — live-formula spreadsheets, sequenced storyboards, flighted plans.
The platform
Everything between the brief and the approved deliverable.
- 01
Model-agnostic routing
Route each client, agent, or task to the model that fits — and when a better one ships, just point the routing at it. Bring your own provider keys, per agency or per client. No lock-in.
- 02
Budgets and metering
Every generation is metered and attributed — user, task, agent, client, campaign — with monthly budget caps that stop the spend before it happens.
- 03
Durable storage and backups
Run on Postgres with persistent uploads, daily off-box backups, and a restore runbook. Agencies can point working files at S3, Google Cloud Storage, Google Drive shared drives, or SFTP — or keep the whole installation, database and files, on infrastructure they run.
- 04
Delivery in real formats
Native Google Docs, Slides, and Sheets, plus PDF, PPTX, XLSX, and PSD with editable text layers — deliverables land in the files your clients already open.
- 05
Media planning and AV
Channel strategy, flighting plans, personas, competitive analyses, and content calendars — with sourced figures and live formulas. The AI drafts the plan; it cannot invent a rate.
- 06
Scripts, treatments, and storyboards
Move from an approved idea into AV scripts, treatments, frame-by-frame storyboards, and reviewable boards without losing the client context or approval lineage.
- 07
Developer ecosystem
A versioned public API, OpenAPI references, SDKs, OAuth, scoped keys, signed webhooks, and usage metering let client portals, DAMs, dashboards, and automation tools plug into the same core.
- 08
White-label and multilingual
The installation carries your agency’s name, logo, and accent colour. English and French ship built-in; more languages upload at runtime — no redeploy.
- 09
Try it without keys
Mock mode runs the whole workflow — intake to review — with no AI keys and no client data, so you can evaluate the fit before connecting a model.
One core, every surface
One workflow, wherever the decision happens.
The web workspace, public API, and mobile companion share permissions, tenant isolation, budgets, context, and the same human review gate.
- Live today
The web workspace
The full workflow — intake, generation, review, delivery — on your agency’s own subdomain, white-labelled and multilingual.
- Live · opt-in
Public API v1
Agencies decide exactly which clients and capabilities each integration can reach — scoped, revocable keys, per-key rate limits, OAuth, signed webhooks, and OpenAPI 3.1 references.
- Live today
The mobile app
The review gate in your pocket: your “waiting on you” queue, approve or send back with notes, browse work offline, and pair with a QR scan.
One service layer underneath — an approval made on a phone is the same record as an approval made at a desk.
AI behind the scenes
The context builds itself.
Your client work already lives in email. Point Tempera at a mailbox and it turns those threads into each client’s working context — quietly, in the background, before anyone opens a brief.
- 01
It reads what lands
Cc, forward, or bcc client email to a mailbox you own. Tempera reads every new thread and works out what’s client work and what’s noise.
- 02
It finds the right client
Each thread is matched to the client it belongs to — your own address and domain rules decide, and the model only assists; it never overrides them.
- 03
It becomes living context
Confirm once, and the thread files itself as that client’s context. Every follow-up folds in on its own, so each agent works from the latest reality.
The reading, sorting, and filing run on their own. A person still holds the pen — nothing becomes context without a nod.
Quick answers
All questions- Can we define our own task types?
- Yes — agencies author task types freely within their plan's deliverable formats; a custom type runs the full pipeline with no code change.
- Which formats can it deliver?
- Eight: text, documents, decks, social posts, graphics, layered graphics, spreadsheets, and storyboards — plus native Google Docs, Slides, and Sheets.
- Does our data have to leave our infrastructure?
- No — run it hosted or entirely self-hosted at the same price, with storage you point at either way.
- Who can approve client-ready work?
- On a team, an authorized reviewer other than the person who ran the task — Tempera rejects self-approval and records the reviewer, verdict, and reason. Working solo, you record your own verdict; the separation switches on the moment a second person is active.
- Does it replace our existing client workflow?
- No — it works inside Client → Brand → Campaign → Activity, then delivers into the documents, decks, spreadsheets, graphics, and storage your team already uses.
One task, end to end
A custom workflow with its proof attached.
Aurora Coffee / Summer launch
Custom task type
Social postsLaunch social set
Grounded in
Brand guidelines + 12 past verdicts
Working storage
Your S3 bucket
Structured output
Three reviewable options
- 1Cold brew, warm welcomes. Summer opens at Aurora.
- 2It’s summer somewhere — ours starts now.
- Golden hour, iced pour. Meet the new summer menu.
Made by
Jordan Lee
Approved by
Maya K. · a different person
Client-ready
The source, selected option, verdict, owner, and storage location stay attached.
Prove it on one client
See one client move from context to approved work.
Run a real brief through Tempera in mock mode. Follow it from grounded options to a named human verdict and client-ready work.
No AI keys. No card. Nothing reaches a client without review.