AI Generative UI (BytesRaw)
by Fasil https://www.bytesraw.com , Bytesraw https://www.bytesraw.com$ 118.99
AI Generative UI (BytesRaw)
AI Workspace: chat that renders rich, interactive reports over Odoo data
Ask a question.
Get a live report.
AI Generative UI adds the AI Workspace to Odoo: a full-screen, ChatGPT-style canvas where you ask for a report or dashboard over your own data and get back live, interactive UI in a preview panel beside the conversation.
Ask for “sales by salesperson this quarter with a trend” and the agent reads the real numbers with its tools, then designs a report — KPI tiles, charts, decorated tables, record links — that renders as native Odoo UI. Maximize the preview, pop it out to its own tab, or print it to PDF.
The critical part is what the model is not allowed to do. It proposes a small declarative component tree from a fixed vocabulary — never HTML, never JavaScript — and it never supplies the figures. Record-bound nodes only name which records and fields to show; the values are re-read on the server as you.
Reports From a Sentence
Describe what you want to see and get a designed report, not a wall of text. The agent picks a layout archetype that fits the question — executive overview, trend deep-dive, comparison, breakdown, scorecard — and chart types that fit the data shape.
Live, Not a Screenshot
Charts, KPIs and tables bound to a model, domain and grouping are recomputed server-side each time the report is opened. The report stays current as your data changes instead of freezing the numbers at the moment it was generated.
Safe by Construction
Only a fixed component vocabulary is accepted, the tree is capped in size and depth, and every figure is re-read under your own access rights. Buttons run only AI-whitelisted server actions, as you, audited, with a confirmation step for anything destructive.
A reporting canvas
the model cannot break out of.
Twelve capabilities, built on one rule: the AI decides what to show, the server decides what the numbers are.
The AI Workspace
A dedicated full-screen surface at AI → Workspace: your conversation list on the left, streaming chat in the middle, and a live preview panel on the right that you can hide, resize, maximize or pop out. Each conversation is private to you.
A Constrained Component Vocabulary
Reports are built from a fixed set of nodes — text, markdown, card, grid, section, divider, table, list, KPI, chart, badge, progress, sparkline, record reference, image and button. There is no arbitrary HTML or JavaScript anywhere in the pipeline.
Seven Chart Types, Chosen by Data Shape
Bar, horizontal bar, stacked, line, area, pie and donut, with legends, value labels and multiple series. The agent is instructed to match the type to the data — time series to lines, share-of-whole to donuts, rankings to horizontal bars — rather than defaulting to one shape for everything.
Design Tokens, Never Raw CSS
Styling is expressed as whitelisted tokens: accent, size, card variant, number format, alignment, and a report theme (executive, minimal, vibrant or dark). The renderer maps them to the theme’s CSS variables, so a report always looks like Odoo and never like injected markup.
Server-Resolved Figures
Bound tables, charts and KPIs carry a model, a domain and a grouping or field list — not values. Everything is resolved on the server, as the requesting user, with read caps applied. A model that tries to hand you a number it invented simply has no channel to do so.
Clickable Reports
Record reference nodes open the underlying record with the user’s own access deciding what they see, and button nodes run server actions that were explicitly flagged Available to AI — as the user, written to the tool audit log, and confirmation-gated when the action is destructive.
Report Style Skills
Four report styles ship ready to use — Executive Clean, Analyst Dense, Vibrant Marketing and Board Minimal — and any skill can be flagged as a report style. Attach one to a conversation and its instructions, plus any brand-guideline documents it indexes, refine every report generated in that chat.
Pop Out and Print
Any generated report opens as a standalone, printable page in its own browser tab, rendered as plain HTML with bound aggregates re-resolved fresh — so Print / Save as PDF produces a clean document with no chat furniture around it.
Streaming Conversation
Answers stream as they are produced, with the chat prose kept deliberately short — one or two sentences — because the preview panel carries the detail. Conversations are auto-titled, keep a bounded history, and expire on the agent’s own idle window.
Reports From Files, Too
Send an image, PDF or document into the conversation and the agent can work from it, subject to what the connected provider supports.
Works With Any Agent
The Workspace injects the rendering tool, the report-style tool and the read tools into every call, so any agent you pick can drive a report regardless of how its own tool list is configured. A Workspace agent ships with the module, tuned for report design and pre-set for large generations.
Manager Oversight
AI → Configuration → All Workspace Chats gives AI managers a read-only view of every user’s conversations and messages — read-only by record rule, so a manager can never edit or delete someone else’s chat.
Connect an agent.
Open the Workspace.
The shipped agent is pre-tuned for report design; everything after that is styling.
Check the prerequisites
A tested provider under AI → Configuration → AI Providers, and AI Discuss installed (it is a dependency).
Wire the shipped agent
At AI → Agents → Workspace, add a connection line with a model that has a large output budget.
Grant access
Users need AI / User to open the Workspace; AI / Manager additionally unlocks the oversight menu.
Choose a report style
Attach one of the shipped Report Styles to a conversation, or create your own with Is Report Style ticked.
Add your brand guidelines
On a report-style skill, turn on Use RAG and upload your design documents so reports follow them.
Expose actions to buttons
A server action must be flagged Available to AI before a report button can run it. See AI → Tools.
See it in action
Generating a report in the Workspace, then taking it out of Odoo.
Four workflows, end to end
Generate a report, apply your house style, share it outside Odoo, and use it as a way into the records.
Generate Your First Report
Turn a question about your data into an interactive report.
Apply Your House Style
Make every report follow your company’s reporting conventions.
Share a Report Outside Odoo
Get a generated report to someone who will not open your database.
Go From a Report to the Records
Use a report as a starting point rather than an endpoint.
Full transparency
The current limitations, stated plainly.
Reports need a capable model
A report is several data reads followed by one long generation, so a small or heavily quantized model will often produce an invalid or truncated report spec. Give the Workspace agent a strong model on its first connection line and leave Max Tokens high — reports fail more often from a cramped output budget than from anything else. The Workspace raises the agent’s step and time limits for report generation, but those raises are still clamped by the agent’s own caps.
The report tree is deliberately bounded
There is a maximum node count and nesting depth, and read caps on bound tables, charts and KPIs. A request that would need hundreds of rows or groups is truncated rather than rendered in full — ask for an aggregate view instead.
The component vocabulary is fixed
Anything outside the supported nodes, chart types, format tokens and icon set is not renderable. An unrecognised icon is dropped rather than failing the report, but an unsupported node type is rejected.
Printed reports are snapshots
Popped-out and printed reports are static HTML taken at the moment you open them. They re-resolve their bound figures on that request, but they are not live afterwards.
Questions, answered
Quick answers to the questions buyers ask most.
Can the AI show me data I am not allowed to see?
No. It never supplies field values at all — bound nodes name which records and fields to show, and the server re-reads them as you, applying record rules, model access and field-level groups.
Can a generated report contain scripts or tracking?
No. The model proposes a declarative tree from a fixed vocabulary; there is no path for raw HTML, JavaScript, frames or external URLs to reach the renderer. Styling is expressed as whitelisted tokens, not CSS.
Are the numbers live or frozen?
Live, when the agent uses bound nodes — which it is instructed to prefer. Values it derived through its own arithmetic are literal and therefore fixed.
Who can see my Workspace conversations?
Only you, plus AI managers through the read-only oversight menu. A manager can read but never edit or delete someone else’s chat.
Do I have to use the Workspace agent?
No. Any agent you can use works — the Workspace injects the rendering and read tools it needs per call. The shipped agent is simply pre-tuned for the job.
Can the report do something, not just show something?
Yes, through button nodes — but only for server actions explicitly flagged Available to AI. They run with your access rights, land in the tool audit log, and ask for confirmation when they are destructive.
Does this replace Odoo’s own reporting?
No. It is for the ad-hoc questions that do not have a report yet — exploratory analysis, one-off dashboards, quick summaries for a meeting.
Release history
A clean record of every release.
Bytesraw
🛠️ Need help?
Got a question, found a bug, or need a custom enhancement? Our support team responds within 24 hours on business days.
help@bytesraw.com →💼 Custom development
Need this module tailored to your workflow, or a brand-new Odoo solution? Let’s talk about your project.
info@bytesraw.com →| Availability |
Odoo Online
Odoo.sh
On Premise
|
| Odoo Apps Dependencies |
Discuss (mail)
|
| Community Apps Dependencies | Show |
| Lines of code | 11001 |
| Technical Name |
ai_generative_ui_bytesraw |
| License | OPL-1 |
| Website | https://www.bytesraw.com |
Odoo Proprietary License v1.0 This software and associated files (the "Software") may only be used (executed, modified, executed after modifications) if you have purchased a valid license from the authors, typically via Odoo Apps, or if you have received a written agreement from the authors of the Software (see the COPYRIGHT file). You may develop Odoo modules that use the Software as a library (typically by depending on it, importing it and using its resources), but without copying any source code or material from the Software. You may distribute those modules under the license of your choice, provided that this license is compatible with the terms of the Odoo Proprietary License (For example: LGPL, MIT, or proprietary licenses similar to this one). It is forbidden to publish, distribute, sublicense, or sell copies of the Software or modified copies of the Software. The above copyright notice and this permission notice must be included in all copies or substantial portions of the Software. THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
Please log in to comment on this module