We build AI that takes real work off your team.
PeopleSimply is a Toronto technology company. We build AI agents, custom apps and integrations for small and mid-sized businesses, and we train the people who use them. We run our own business on the same kind of agents we build.
People Simply Inc. is a Canadian technology company based in Toronto.
Nav Sivapatham spent two decades in enterprise technology, most recently at Alight Solutions. He has built with AI hands-on for the past three years.
Tell us about the job your team spends too long on. We'll tell you plainly whether AI can help.
| No. | Service | What it looks like | Section |
|---|---|---|---|
| 01 | Custom AI agents | Software that handles one recurring job end to end, like turning a stack of drawings into a draft scope, with a person approving the steps that matter. | |
| 02 | Custom apps | Web and mobile apps built for the way your team works, when off-the-shelf tools don't fit. | |
| 03 | Workflow automation and integrations | Connecting the systems you already pay for, so nobody retypes data from one into another. | |
| 04 | AI training for staff | Live, hands-on courses that teach your team to use ChatGPT, Claude and Copilot on their own work. |
We run PeopleSimply on a team of AI agents.
Before we build agents for anyone else, we run our own business on them. A team of agents works alongside our founder every day. They coordinate the work, write and ship software, prepare documents and invoices for review, triage what comes in, and check that everything still runs.
This is a production system, and it has been for months. The agents are always on and keep lasting memory. They hand work to each other with written briefs and answer to one set of rules that's checked automatically every day.
Every pattern we sell, we ran on ourselves first. The mistakes we made along the way are written into the rules.
← doneAgentbrief →
← doneAgent
- Always on
- Agents run around the clock in their own workspaces. They pick up work as it arrives, from a chat message, a schedule or another agent.
- Persistent memory
- Each agent keeps lasting notes on the business: decisions, people, and how things are done. Tomorrow starts where today ended.
- Agent to agent
- Agents hand work to each other with a written brief. The receiver confirms it, then reports back when it's done or stuck, so nothing gets lost between them.
- Guardrails
- Every agent starts from the same hard rules. Nothing that leaves the business, spends money or can't be undone happens without a person's go. Automated checks run daily to catch drift.
- Across systems
- Agents run on more than one machine and work inside the tools a business already uses.
- People in charge
- Every agent reports what it did, and what it didn't do.
- Monitored
- More than a hundred automated checks run every morning and on every restart. A new failure reaches the founder's phone.
- Recoverable
- Memory and work are backed up every hour to a separate machine.
| Role | What it does |
|---|---|
| Chief of staff | Coordinates the other agents, tracks the founder's priorities and commitments, and reports what's done and what isn't. |
| Engineer | Designs, builds, tests and deploys software: web and iPhone apps, integrations, and the agents themselves. |
| Client operations | Runs inside a client's business on their own hardware, works in their project and CRM systems, and prepares the reports their team used to build by hand. |
| Scheduled work | Invoicing, lead triage, backups and health checks run on their own schedule and speak up when something is off. |
Custom AI agents
An agent is software that reads, decides and drafts on its own, with a person approving anything that matters. We build each one around a single recurring job, like the status report someone rebuilds every Monday, or the quote that starts from a stack of drawings.
Every agent we hand over comes with a run note. It says what starts the agent, who owns it, where a person signs off, and how to switch it off.
When one agent isn't enough, we build a team of them, the way we run our own business.
We build on Claude and OpenAI models.
- Cut the time it takes a designer to build a design package.
- Automated part of a project coordinator's weekly work.
- Trigger
- What starts it: a schedule, an email, a message.
- Owner
- The person it answers to.
- Sign-off
- Where a person approves before it acts.
- Memory
- What it keeps between runs, and where.
- Guardrails
- What it may never do on its own.
- Health checks
- How we know it's still working, and who hears when it isn't.
- Report
- What it did and what it didn't, every run.
- Off switch
- How to stop it in one step.
Custom apps
Some work doesn't fit a product off the shelf. The schedule lives in a spreadsheet only one person understands, or the team pays for three tools that each do part of one job. We build the app that fits how your business runs.
That can be a web app your staff open in any browser, or a native iPhone or Android app for people who work away from a desk. AI goes in where it saves someone time, and nowhere else.
- Web apps
- Laravel · Vue
- iPhone and iPad
- Swift · SwiftUI
- Android
- Kotlin · Jetpack Compose
- AI features
- Claude · OpenAI
Workflow automation and integrations
Most businesses already pay for good tools that don't talk to each other. So someone copies data from one into the next, every day, and each copy is a chance for a mistake.
We connect the systems you have so the data moves on its own. When a plain automation will do the job, we'll say so and build that instead of an agent. It's cheaper to run and easier to trust.
If a system has an API, we can usually connect it. Where there's no ready-made connector, we build one, including connectors that let AI agents work inside the system directly (MCP servers). We built our own for Zoho Projects when none was available.
- Automated a weekly status report that covers many sites at once.
- Systems we've connected
- Zoho Projects, CRM and WorkDrive · Microsoft 365 · Gmail · Google Drive · Telegram · Otter
| Typical job, by hand | Once it's automated |
|---|---|
| Someone pulls figures from every active site into a weekly report | The report builds itself each week from the project system |
| Orders arrive by email and get retyped into the job system | Each order lands in the job system with the email attached |
| A new hire's details get entered into four different tools | One form fills in the rest |
AI training for your staff
We teach from daily practice. Our own business runs on these tools, so the course comes from what works in real jobs, and we revise it as the tools change.
It covers ChatGPT, Claude and Microsoft Copilot, from writing a first good prompt to building an agent with a person signing off on its actions. Every exercise uses your team's own work, so people leave with something they'll use the next day.
Sessions are live and instructor-led, on-site or online. Each module ends with a practical exercise scored against a written rubric.
- Modules
- 6
- Format
- Live, on-site or online
- Tools
- ChatGPT · Claude · Copilot
| No. | Module | Afterwards, people can |
|---|---|---|
| 1 | Prompting and task designChatGPT, Claude, Microsoft Copilot | Write prompts whose output meets a clear standard, fix one that fails, and pick the right assistant for the task. |
| 2 | Verifying AI output and handling data safely | Check AI work against its sources and catch the errors, and know which business data may go into an AI tool. |
| 3 | AI in Microsoft 365 | Draft email replies, turn a meeting into decisions and owners, and condense a long document into a one-page brief. |
| 4 | AI-assisted data analysis | Clean a raw export, answer a business question with a summary or pivot, and check the figure by hand. |
| 5 | Building AI agents and automations | Build an assistant and a no-code automation that run correctly on test cases, and set a person to sign off on each action. |
| 6 | Applied capstone | Take one real recurring task from their own job, do it end to end with AI, and record the time it took before and after. |
We keep asking what we can do better.
AI changes month to month. New models, new features and better ways of working arrive faster than any one project, so staying current is a standing part of our week.
We read the research and the release notes, and we try new models and features on our own work first. We keep what proves itself. When something goes wrong, or a better way turns up, we write the lesson into the rules our agents and our team work from, so the same mistake doesn't happen twice.
Our clients get the benefit. What we learned last month is already in what we build and teach this month.
| No. | Step | What it means |
|---|---|---|
| 1 | Watch | New models, new features and published best practice. |
| 2 | Try | On our own operation first. |
| 3 | Measure | Against the way the work was done before. |
| 4 | Keep or drop | And write down why. |
| 5 | Write it down | The rule changes, and every agent starts from the new one. |
Tell us about the job your team spends too long on.
A few lines is enough: what the work is, who does it now, and how often it comes round.
Email info@peoplesimply.com