← All projects
Case study · AI agents · SaaS · Live
VYNORO

An AI team for small agencies that plans the work, shows its sources, and asks before anything goes out.

A one-to-ten person agency can't hire a researcher, an analyst and a copywriter for every prospecting list or client report. Chatbots help, but their answers come unchecked, with contacts and numbers that are sometimes invented. VYNORO gives the owner a team of AI specialists led by Vyn: describe the job, Vyn plans it, the team works with real tools, every claim links to its evidence, and nothing is sent without the owner's approval.
RoleSolo: product, design, full stack, AI runtime
Timeline2026
TypeMy own SaaS product, live
StackNext.js 16, React 19, Supabase, Gemini on Vertex AI, Three.js
vynoroai.com/dashboard
VYNORO workspace dashboard
7
AI teammates, led by Vyn
9
agent tools, from Maps to PageSpeed
25
migrations, RLS on every table
$1
default budget per mission

Overview.

Who it is for, what goes wrong today, what VYNORO does.
The audience

Agencies of one to ten people

Web, marketing and design studios where the owner sells, delivers and does the admin. The screens on this page come from a workspace for my own agency, Barmjli, running real missions on public data.

The problem

Too many jobs, no one to hand them to

Finding prospects, auditing their sites, writing first messages and preparing client reports all take hours. Generic AI chat helps with each piece but nobody checks the result, and an invented email or number reaches a client.

The product

A team that works and shows its evidence

Vyn plans each request and hands steps to specialists with real tools. Code and Vyn check every output, the report cites its sources, and drafts wait for the owner. The team learns from every correction.

Walkthrough.

One real mission, start to finish: find restaurants in Milan with weak websites and draft a first message for each.
01 · Workspace

One screen that says what needs the owner

The owner of a small agency opens VYNORO and sees which agents are working, what is in progress, what needs a decision and what is done. A workspace holds one business, so an agency can run its own brand and its clients side by side.

vynoroai.com/dashboard
VYNORO workspace dashboard
Agents at work, missions in progress, items that need the owner, and the team
  • 1
    Attention firstMissions waiting for a review or an answer sit at the top and in the sidebar badge, with a link straight to the decision.
  • 2
    A company profile the team readsWhat the business sells, its prices, tone and region live in the workspace profile and go into every prompt, so the work sounds like the business.
  • 3
    Plans and creditsEach workspace shows its plan and credits used. Limits are checked on the server before a mission starts, not only in the UI.
02 · The office

A 3D office where you can watch the team work

Each specialist has a desk in a sunlit office built with React Three Fiber. Agents light up while they work, and the mission board on the wall shows what is in progress, what needs the owner and what is done.

vynoroai.com/office
3D office with the AI specialists at their desks
Six specialists and Vyn, with live status from the mission board
vynoroai.com/office
Vyn close-up in the 3D office
Select an agent to see what it is working on
  • 1
    Real data, no demo scriptThe office only shows what the mission backend publishes. With no missions running, the team is idle.
  • 2
    Live statusMission steps and events stream in through Supabase Realtime, so an agent starts moving the moment its step begins.
  • 3
    Camera moves with careCamera paths and agent animations have their own tests, so a change to the office doesn't break the view.
03 · Ask Vyn

Describe the job once, and Vyn turns it into a plan

The owner writes what they need in plain words. Vyn, the chief of staff, decides whether it can answer on the spot, needs one clarifying question, or needs a plan. A plan names the specialist for each step and what that step must hand over.

vynoroai.com/dashboard
Talk to Vyn with a three-step plan
A Milan prospecting request split into research, audit and outreach steps
  • 1
    Quick answers stay quickA question, a fact or one short text is answered directly in a few seconds, without starting a mission.
  • 2
    One question, not tenWhen something essential is missing, Vyn asks the owner one question, then plans.
  • 3
    Steps run in parallelSteps whose inputs are ready run together, up to three at a time, and each one hands its output to the next.
04 · Report

A report where every claim points to its source

When a mission ends, Vyn writes a briefing and the Presenter lays out the result: what was found, the evidence, the strategy and the next action. Every mark in the report links to the page, search or calculation it came from.

vynoroai.com/report
Mission report headline
The headline, the request, sources checked and Vyn's corrections
vynoroai.com/report
Prospects with technical audit findings
Each prospect with the flaw that was measured on its site
vynoroai.com/report
Outreach schedule and strategic angle
Outreach schedule and the angle behind it
  • 1
    Sources listedEvery page read, Maps search and site check is listed beside the report, grouped by website.
  • 2
    Numbers must exist in the evidenceA report or chart block whose number doesn't appear in the source material is dropped before it is shown.
  • 3
    Export to OfficeResults export to PowerPoint, Excel and Word, or a PDF, so they go straight to a client.
05 · Approval

Nothing is sent until the owner finishes the review

Drafts from the Outreach specialist land in a review queue. The owner reads each one, edits it, approves it or sends it back with a note, and nothing leaves the workspace before the review is finished.

vynoroai.com/work/review
Email review queue
Five drafted emails, each tied to a flaw found on the prospect's site
  • 1
    Edits are keptChanges to a draft are saved, and approving the rest takes one click once the owner is satisfied.
  • 2
    Lessons from correctionsA rejection with a note or an edit becomes a short lesson, such as how to sign emails, that goes into later missions.
  • 3
    Contacts must be realAn email address counts only if it appeared on a page the team actually opened. Invented contacts are caught by code.
06 · Arabic

The same team in Arabic, right to left

Ask in Arabic and the plan, the work and the report come back in Arabic, laid out right to left. Here the team looked for restaurants in Algiers and reported what each one's website was missing.

vynoroai.com/report
Arabic mission report
An Arabic report on restaurants in Algiers, with sources
  • 1
    Written in ArabicAgents write in the language of the request, and the report layout mirrors for right-to-left.
  • 2
    Mixed sourcesSources in French, Arabic and English sit in the same evidence list.
  • 3
    Same checksThe code checks and Vyn's review run the same way in every language.
Routines on a schedule
Knowledge library
Brand kit
PowerPoint, Excel and Word export
Plans and credits
Admin console and audit log
Arabic RTL
Voice input

The team.

Vyn manages, the specialists do the work, and every step goes through the same loop.
1 · SortVyn decides in about a second whether to answer now, ask one question, or plan.
2 · ActSpecialists call their tools. Each result becomes citable evidence.
3 · CheckCode looks for invented emails, missing evidence and unsupported numbers.
4 · ReviewVyn accepts, asks for a revision (up to two), or escalates to the owner.
5 · DeliverReport, files and drafts. Drafts wait for the owner's approval.
Vyn · Chief of staff
  • Sorts each request
  • Plans the steps
  • Reviews every output
  • Writes the briefing
Researcher · Finds
  • Web search
  • Google Maps places
  • Reads pages and PDFs
Analyst · Measures
  • PageSpeed and site checks
  • Exact calculations in code
  • Compares options
Writer · Writes
  • Posts, pages and replies
  • Follows the brand tone
  • Learns from edits
Designer · Shows
  • Images
  • Presentation layouts
  • Brand kit colours and fonts
Outreach · Drafts
  • Personal first messages
  • Follow-ups
  • Never sends on its own
Operations · Handles the rest
  • Any task no other role fits
  • Lists, files and data

Architecture.

One Next.js app, one Supabase project, and an AI runtime that runs on the server.
Client
Next.js 16 App RouterReact 19 · TypeScript · Tailwind v4 · framer-motion · 34 pages · English and Arabic
3D officethree · React Three Fiber · drei · agents driven by live mission events
Mission reportBriefing, Presenter blocks and charts · export to pptx, xlsx, docx and PDF
→
Server
Mission runtimeSort → plan → specialists with tools → code checks → Vyn's review → deliverables
Background runsOwner session on the server · 4-minute lease · watchdog every minute · resume from the last step
API routes and scheduler37 route handlers · routines claimed with FOR UPDATE SKIP LOCKED · plan limits checked server-side
→
Data and models
SupabasePostgres with owner-only RLS · Realtime for live missions · private Storage for deliverables · Auth
Gemini on Vertex AIPro for Vyn, Flash for specialists, Flash-Lite as fallback · retries, timeouts and cost per call
Google toolsSearch and Maps grounding · PageSpeed · code execution · image generation
About 40,000 lines of TypeScript. Prompts come in four versioned layers: a constitution every agent follows, one template per role, the workspace profile, and the task with teammates' outputs.

Engineering deep dives.

Six decisions that make an AI team safe to trust, each with what it cost.

1. Code checks before AI review

Problem
An AI reviewer tends to wave through the most damaging mistakes: an email address that looks right, or a load time nobody measured.
Approach
Before Vyn reviews anything, deterministic checks run on the structured output. Every cited evidence id must exist. Every email must appear on a page the team opened, not just in a search snippet. Every measurement-like number must appear in the evidence or in the owner's own profile. Errors go back to the specialist for one short rewrite, without an AI call to find them.
Trade-off
Regex checks can flag a legitimate number written differently from the source. The owner's own facts count as known to keep that rare.
Lives in: src/lib/ai/runtime/checks.ts

2. Missions that outlive the browser tab

Problem
A prospecting mission can take minutes. An owner who closes the laptop shouldn't lose the work, and a deploy in the middle shouldn't either.
Approach
Missions run on the server, signed in as the owner with their own server session, so row-level security still applies. A run holds a 4-minute lease that every AI call renews. A watchdog checks every minute and continues a dead run from the last finished step. After three automatic resumes, or a run dead for 30 minutes, the mission fails with a Retry button.
Trade-off
On serverless hosts the watchdog needs a cron call every minute, and long runs yield between steps after 9 minutes.
Lives in: runtime/background.ts · /api/cron/missions

3. Nothing leaves without the owner

Problem
An AI that emails a restaurant on its own, with a wrong detail, damages the agency's name. Small agencies live on their reputation.
Approach
Outreach only drafts. Drafts become approval items on the mission, the owner reviews them one by one, and nothing is sent until the review is finished. Edits and rejections with a note are turned into lessons stored per workspace and added to later prompts.
Trade-off
The owner must spend a few minutes reviewing. That is the product's position, not a limitation to remove.
Lives in: runtime/mission.ts · runtime/lessons.ts

4. Owner-only data, enforced in Postgres

Problem
Workspaces hold client lists, prices and drafts. A query that forgets a filter must not leak another owner's data.
Approach
All 25 migrations keep row-level security on, with owner-only policies on every table. The AI runtime works as the signed-in owner, not with the service role, so an agent can't read what the owner can't. Files sit in a private Storage bucket under the owner's id and are served with ten-minute signed URLs.
Trade-off
Background runs need a real owner session created on the server, which is more moving parts than a service key.
Lives in: supabase/migrations · runtime/background.ts

5. Every AI call has a price tag

Problem
Agents that loop with tools can burn money quietly. A small agency needs a cost it can predict.
Approach
Every Gemini call is logged with model, tokens, estimated cost and latency. Missions start with a 1 USD budget and pause at the limit until the owner continues. Models are chosen per tier in one file: Pro for Vyn's reasoning and reviews, Flash for specialists, Flash-Lite as the fast fallback, each with retries and timeouts.
Trade-off
Costs are estimated from token counts, so they track the provider's bill closely but aren't the invoice.
Lives in: src/lib/ai/models.ts · gemini.ts · ai_calls

6. Tools with guardrails

Problem
Agents read the open web. A page can try to reach internal addresses, or hide instructions meant for the model.
Approach
Which role may call which tool, and how often, is enforced in a registry. fetch_page only allows http(s) and public IP addresses, re-checks every redirect hop, and caps size and time. Fetched text is wrapped as untrusted input in the prompt, and every tool result gets an evidence id like E2.3 that the report must cite.
Trade-off
Tight limits mean some heavy pages are read only in part.
Lives in: src/lib/ai/tools/registry.ts · fetch-page.ts