Order receivedExample
Hello, please find attached our order no. 4471: 3 items, delivery requested on the 18th. Could you confirm the lead times? Have a good evening.
AI agents, automations and custom tools for SMBs
AI agents that answer, follow up and enter data so your teams don’t have to.
One agent, one task, one result. It works in your tools, around the clock, and hands over to a human as soon as something falls outside its scope. Built, run and maintained by Mynos.
At Toshify, the onboarding team went from 3 people to 1.
TRUSTED BY
Y. MessaoudiOrthodontist
What we do
An agent when it takes thinking. A workflow when executing is enough. A custom tool when nothing exists.
We start from your process and pick the simplest solution that gets the job done. An agent only when it is worth it; otherwise, an automation, at no extra cost.
- Our core business
AI agents
When the work means understanding a request, deciding or getting several tools to work together, an agent takes over, configured for your rules.
- 6 agents ready to configure
- Custom agents
- Plugged into your tools
- The quick win
Automations
When the task follows a set path, it needs no AI: a workflow runs it, at no extra cost.
- n8n workflows
- Sync between tools
- Automatic reports
- The foundations
Tools and platforms
Your data is scattered, or the tool you need doesn't exist yet? We build your internal interfaces, custom CRMs and dashboards.
- Internal platforms
- Dashboards
- APIs and connectors
Example journey
From purchase order to invoice, with no re-keying.
One evening at 9:07 pm, a client sends their order as a PDF. Nobody is at the office. Here is how the agents take over, step by step.
Example · Industrial supplies distributor, 35 employeesWhen9:07 pm
- 01Received
The order comes in at 9:07 pm, by email, as a PDF, outside office hours.
- 02Reading and checks
Hugo reads the PDF, extracts the items, quantities and date, then checks them against the negotiated price list and the stock. He flags a price gap on line 2.
- 03Approval
At 8 am, sales admin approves the gap in one click. The agent hands over as soon as a decision falls outside its scope.
- 04Entry and confirmation
Hugo enters the order in the ERP and sends the client the order confirmation with the delivery dates.
- 05Follow-up and invoice
Nora answers the client asking where the order stands. Sam invoices on shipping and follows up at the right time.
Extracted data
- ClientExisting account, negotiated prices
- OrderNo. 4471, 3 lines
- Requested deliveryThe 18th
- Stock2 lines available, 1 being restocked
- Gap foundPrice of line 2
- ActionSales admin approval requested
Hover over a value to see where it comes from.
Example journey
From purchase order to invoice, with no re-keying.
One evening at 9:07 pm, a client sends their order as a PDF. Nobody is at the office. Here is how the agents take over, step by step.
Example · Industrial supplies distributor, 35 employees01/05 · 9:07 pmReceived
The order comes in at 9:07 pm, by email, as a PDF, outside office hours.
Order receivedExample
Hello, please find attached our order no. 4471: 3 items, delivery requested on the 18th. Could you confirm the lead times? Have a good evening.
Example journey
From purchase order to invoice, with no re-keying.
One evening at 9:07 pm, a client sends their order as a PDF. Nobody is at the office. Here is how the agents take over, step by step.
Example · Industrial supplies distributor, 35 employees- ✉01 · 9:07 pmReceived
The order comes in at 9:07 pm, by email, as a PDF, outside office hours.
Order receivedExampleHello, please find attached our order no. 4471: 3 items, delivery requested on the 18th. Could you confirm the lead times? Have a good evening.
02 · 9:08 pmReading and checksHugo reads the PDF, extracts the items, quantities and date, then checks them against the negotiated price list and the stock. He flags a price gap on line 2.
Alert from HugoExampleOrder no. 4471 from Sophie Marchal, 3 lines. Line 2: €18.40 per unit on the purchase order, €16.90 on the negotiated price list. Lines 1 and 2 in stock, line 3 being restocked (available on the 21st). Which price should I apply?
Negotiated priceOrder priceExtracted data- ClientExisting account, negotiated prices
- OrderNo. 4471, 3 lines
- Requested deliveryThe 18th
- Stock2 lines available, 1 being restocked
- Gap foundPrice of line 2
- ActionSales admin approval requested
Hover over a value to see where it comes from.
- 03 · 8:02 amApproval
At 8 am, sales admin approves the gap in one click. The agent hands over as soon as a decision falls outside its scope.
Approved by sales adminExampleOrder no. 4471 from Sophie Marchal, 3 lines. Line 2: €18.40 per unit on the purchase order, €16.90 on the negotiated price list. Lines 1 and 2 in stock, line 3 being restocked (available on the 21st). Which price should I apply?
✓ Negotiated priceOrder priceApproved by Claire at 8:02 am
- 04 · 8:04 amEntry and confirmation
Hugo enters the order in the ERP and sends the client the order confirmation with the delivery dates.
Confirmation sentExampleHello Sophie, your order no. 4471 is confirmed. Lines 1 and 2: delivery on the 18th. Line 3: delivery on the 21st, while it is restocked. Prices are those of your negotiated price list. Have a good day.
05 · The 18thFollow-up and invoiceNora answers the client asking where the order stands. Sam invoices on shipping and follows up at the right time.
WhatsApp · NoraExample- SophieHello, can line 3 of order 4471 arrive before the 21st?
- NoraHello Sophie, line 3 is being restocked: delivery is still planned for the 21st. I will let you know as soon as it leaves the warehouse.
Invoice sent on shipping · Sam
The agents
One AI agent per job, configured for your business.
Front desk, recruiting, sales, orders, invoices: each one takes over a precise task and hands over to a human as soon as it falls outside its scope. Six agents ready to deploy; for a more specific need, we build or configure the agent for you.
- AttractMaya
- WelcomeNora
- ConvinceLéo
- EnterHugo
- Get paidSam
- RecruitIris
Maya
Content officer
≈ 6 h 15 saved per weekEstimate for 5 posts a week (1.5 h of creation each, versus 15 min of review with the agent).
Creates the visuals, writes the posts and keeps your LinkedIn and Instagram calendar.
- LinkedIn posts and carousels
- Visuals and short videos
- Editorial calendar
Nora
Front desk officer
Replies in under 30 s, 24/7In service at Guardería Jardín San José, 200+ messages from parents a day.
Replies on WhatsApp and by message, qualifies requests and books appointments in your calendar.
- 24/7 replies to messages
- Request qualification
- Appointment booking
Léo
Sales assistant
≈ 9 hours saved per weekEstimate for 80 quotes and 120 calls a month (3 five-minute follow-ups per quote, 10 min of CRM entry per call).
Follows up open quotes, updates the CRM after every call and flags the deals that need attention.
- Personalised quote follow-ups
- Call notes in the CRM
- Qualified meetings in your calendar
Hugo
Sales admin assistant
≈ 12 hours less data entry per weekEstimate for 800 orders a month, 4 min of entry each.
Reads orders received by email or as PDFs, enters them in your ERP and flags the ones that raise questions.
- Reading purchase orders
- Entry into the ERP
- Anomaly alerts
Sam
Accounting assistant
≈ 7 h 45 saved per weekEstimate for 300 invoices a month, 90 of them late (2 ten-minute reminders each), and 4 h of accounting export a month.
Tracks invoices, chases late payments and prepares the numbers for your accountant.
- Invoice tracking and reminders
- Financial dashboards
- Accounting exports
Iris
Recruiting officer
247 candidates qualified in 24 hIn service at Toshify (Buenos Aires): 95% of candidates handled with no human intervention.
Replies to candidates, checks their documents, chases the missing ones and passes on complete files.
- Application screening
- Document checks
- Automatic follow-ups
Custom
Your processes are unique?
We design the agent that follows your rules, speaks your language and plugs into your tools.
Talk about my agentCase studies
Some of our work
A selection of projects delivered for our clients, in France and abroad.
Toshify
Buenos Aires · An AI onboarding assistant for Argentine drivers who want to join Cabify. Qualification, document checks and activation, 24/7 on WhatsApp.
Read the caseIllustrative data
Heex Technologies
A sales control dashboard: automatic CRM data sync, pipeline and priority deals, AI scoring of call transcripts against the sales script.
Read the caseGuardería Jardín San José
Trujillo, Peru. Bilingual website and WhatsApp agent for parents. 200+ messages a day, replies in under 30 s, 24/7.
Read the caseTaximoove
Complete taxi platform: web app, iOS and Android apps, payment, back office. 847 rides a day, 4 min 12 s average wait.
Read the caseThe team
Three partners on every project: operations, tech and strategy.
Operations & financeJean-Philippe Martial
Nearly ten years across operations (Bolt, Meero, Stilla Technologies) and corporate finance. He spots where your processes lose time and puts a number on the gain before automating them.
LinkedIn
Advanced technologyMalik Fahem
Senior engineer in AI, data architecture and product. He picks the building blocks (models, hosting) and ensures robust, maintainable deliverables.
LinkedIn
Strategy & businessTimothée Collange
Graduate of X-HEC Entrepreneurs and Berkeley, former COO of Alice & Bob (quantum computing) and founder of several startups. He frames every project and checks that it starts with a clear ROI.
LinkedIn
Method
From the first call to an agent in production.
For a first agent on a focused task, allow 2 to 4 weeks.
- Step 1/5
Analysis
We map your processes and find the tasks where an agent pays off the most.
Deliverables- Map of your processes
- Estimated gains task by task
- Choice of the first agent
- Step 2/5
Design
We define the agent's role: its scope, tone, rules and limits.
Deliverables- Agent specification
- Business rules and edge cases
- Testable mock-up
- Step 3/5
Integration
We plug the agent into your tools: CRM, messaging, calendar, accounting software.
Deliverables- Connections to your tools
- Tests on your real data
- Security and GDPR review
- Step 4/5
Deployment
Gradual go-live, with a human in the loop at the start.
Deliverables- Supervised go-live
- Onboarding for your teams
- Activity dashboard
- Step 5/5
Optimisation
We measure, adjust, then extend the agent to other tasks.
Deliverables- Monthly report
- Continuous improvements
- New agents if needed
Method
From the first call to an agent in production.
For a first agent on a focused task, allow 2 to 4 weeks.
Analysis
We map your processes and find the tasks where an agent pays off the most.
Deliverables- Map of your processes
- Estimated gains task by task
- Choice of the first agent
Design
We define the agent's role: its scope, tone, rules and limits.
Deliverables- Agent specification
- Business rules and edge cases
- Testable mock-up
Integration
We plug the agent into your tools: CRM, messaging, calendar, accounting software.
Deliverables- Connections to your tools
- Tests on your real data
- Security and GDPR review
Deployment
Gradual go-live, with a human in the loop at the start.
Deliverables- Supervised go-live
- Onboarding for your teams
- Activity dashboard
Optimisation
We measure, adjust, then extend the agent to other tasks.
Deliverables- Monthly report
- Continuous improvements
- New agents if needed
Resources
What we learned by delivering.
- 7 min read
Why 80% of AI projects fail (and how to avoid being one of them)
Read the article - 8 min read
Scoring 100% of sales calls with AI: what really changes
Read the article - 10 min read
n8n vs Make vs Zapier: the honest comparison after a dozen projects shipped
Read the article
Frequently asked questions
What business leaders ask us.
Software that carries out a task end to end inside your tools: it reads a request, understands it, acts (replies, updates the CRM, sends a quote) and reports back. Unlike a simple chatbot, it does not just answer: it does the work, within the rules you set.
No. When a task follows a set path, an automation is enough and costs less. An agent is justified when the work means understanding a request, deciding or getting several tools to work together. We suggest an agent when it is worth it, not before.
On a focused task, allow 2 to 4 weeks. On a more complex case, such as a custom onboarding agent, allow 8 weeks. On average, the return on investment comes in under two months.
Our projects start from about €800 per month: we build the solution, run it and maintain it. The price depends on the scope, from a single automation to a custom workflow with several agents, and grows with its complexity. Business platforms and tools are quoted at a fixed price, after the assessment, with the expected gain alongside.
No. They take over repetitive tasks (follow-ups, data entry, sorting, first replies) so your people can focus on what needs judgement. A human stays in charge: the agent hands over as soon as a situation is outside its scope.
Yes, that's the idea: it plugs into what you already use (CRM, email, WhatsApp, calendar, accounting software, ERP) through their APIs or n8n and Make workflows. No need to change tools.
Your data is stored in Europe and encrypted. We sign a confidentiality agreement and a data processing agreement (DPA) with every client, and we never reuse client data to train models.
They are the agents we configure most often. Each agent is set up for your business: your rules, your tone, your tools. And if your task isn't on the list, we build the agent that does it.
Yes. We work in French, English and Spanish. Several of our clients are international: Guardería Jardín San José in Peru (Trujillo), Toshify in Argentina (Buenos Aires), and other remote projects for European clients.
Next step
In 30 minutes, we pinpoint what an agent can take over for you.
We review your processes and give you the figure, task by task. Free, no commitment.