ThriveAI vs building it yourself with Claude Code
Claude Code lets a developer on your staff build AI tools in-house. ThriveAI builds a private AI system on your company’s own data and works hands on with your team. This page compares the two for a smaller manufacturer or distributor in Ontario or Quebec, and it says plainly when doing it yourself is the better choice.
The short answer
- Do it yourself with Claude Code if a developer on your staff has time to build, host and maintain the system, and your data can be processed outside Canada. You pay for a Claude plan and its usage, and the know-how stays in your company.
- Work with ThriveAI if nobody on your payroll writes software, or you want a system set up for you on a server in Canada that only your company uses. ThriveAI builds the system with your team and starts with a working prototype for one job, at a fixed price.
Claude and the other AI tools on this page bill in US dollars. Facts about them were checked on September 23, 2026, and wages are in Canadian dollars.
Side-by-side comparison
Each topic sets what you get from ThriveAI beside what the same job asks of you when you do it yourself with Claude Code. Each fact about Claude Code links to its source.
| ThriveAI | Do it yourself | |
|---|---|---|
| What you get | A private AI system built on your ERP, email, drawings and spreadsheets. It answers questions with the source of each number and drafts routine work, such as order entries and purchase orders, for someone on your team to approve. | An agentic coding tool that reads code, edits files and runs commands. The system your developer builds with it is up to them. |
| Who does the work | ThriveAI, working on site with the people who do the job. Derik Lawlis, the founder, leads the project and stays close to the build. | A developer on your payroll. Anthropic says the person using Claude Code is responsible for reviewing each proposed change and command. |
| Tool cost | Every project is quoted on its own, and no rates are published. | A Claude plan for each developer, charged per month on Pro and Max or per seat on Team, plus usage past the plan at API rates. Claude’s pricing page lists the rates. |
| People cost | Time from the people who know the job, to shape the tools and approve what they draft. | A software engineer’s pay. The Ottawa median is $56.73 an hour. |
| First commitment | Stage one: a working prototype for one job, at a fixed price, in weeks. You keep it if you stop there. | A monthly plan on Pro, Max or Team. Enterprise is billed annually with a 20-seat minimum. |
| Where your records are stored | The platform ThriveAI builds on is designed to keep them on a server in Canada that only your company uses. | Where your developer hosts the system. Anthropic’s own API stores workspace data in the US only. |
| Where the model runs | You choose a model on your server, or a hosted model under a written no-retention agreement. A hosted model may process requests outside Canada. | Anthropic’s API processes requests in the US or globally. On Amazon Bedrock from Canada, Opus 5.5 and Sonnet 5 run only through cross-Region profiles. |
| Running it every day | ThriveAI stays involved after launch and keeps adding to the tools as your business changes. | If you host it yourself, your team handles saved state, monitoring, credentials, scaling and cost tracking, according to Anthropic’s hosting guide. Anthropic’s Managed Agents can host the agent loop instead. |
| Who approves what it does | Someone on your team approves every draft before anything is sent or saved in your systems. | Whatever your developer builds in. Claude Code itself only has the permissions you grant it. |
| If you stop | You keep your data, and everything the system recorded exports in full in a documented format. | You keep the code your developer wrote. On self-serve Enterprise, seats cannot be removed during the annual term. |
| Best fit | A company without a developer on staff that wants one job working first. | A company with a developer who has time to build and maintain the system. |
For more about the company and how a project runs, see About ThriveAI.
What doing it yourself costs
A do-it-yourself build has two costs you can look up: the Claude plan with its usage, and the pay of the person who builds with it.
The Claude plan
Claude Code is included in every paid Claude plan and is not in the Free plan. It shares one usage pool with Claude chat, and heavy users can switch to pay-as-you-go API credits.
| Plan | How it charges | What to know |
|---|---|---|
| Free | No charge | Claude Code is not included. |
| Pro | Per month, or an annual plan paid up front | Claude Code and Claude chat draw on the same usage limits. |
| Max 5x | Per month | Five times Pro usage, billed monthly only. |
| Max 20x | Per month | Twenty times Pro usage, billed monthly only. |
| Team, standard seat | Per seat per month, billed monthly or annually | For 2 to 150 seats. Credit card only, with no invoicing or net terms. |
| Team, premium seat | Per seat per month, billed monthly or annually | Five times the usage of a standard seat. |
| Enterprise | Per seat per month, billed annually | At least 20 seats. The seat includes no usage, and every token is billed at API rates. Self-serve Enterprise bills in US dollars only. |
Sources: Anthropic’s pricing page, its Max plan article and its Enterprise seat article, checked on September 23, 2026.
The smallest Enterprise contract is 20 seats for a year, and usage is billed on top of the seats.
Usage past the plan
Anthropic’s cost guide reports what Claude Code costs per developer per active day across enterprise deployments, and how widely that varies between users. Those averages measure a developer at work in Claude Code, and they do not cover a finished system that your team uses every day.
A finished system that calls Claude through the API pays per million tokens, with one rate for input and another for output, and the rate depends on the model: Haiku, Sonnet, Opus or Fable. Batch processing costs less than the list price, and US-only processing costs more, according to Claude’s pricing page. For the rates of each model, see Claude API pricing.
Anthropic announced a change to how Agent SDK usage is billed on Claude plans, and it paused that change on June 15, 2026. Agent SDK usage still draws from your plan’s limits, so read the terms again before you set a budget.
Other AI coding tools charge the same way. GitHub Copilot Business charges per seat per month, Cursor Teams charges per user per month, and OpenAI Codex comes with ChatGPT Business, which charges per user per month.
The person who builds it
Claude Code is a tool for people who write software. It works in the terminal, a code editor, a desktop app and the browser, where it reads a codebase, edits files and runs commands. Anthropic’s security guide says you are responsible for reviewing its proposed code and commands before you approve them. A company without a developer has nobody to do that review.
Developers use these tools widely, and many distrust what the tools produce. In the Stack Overflow Developer Survey 2025, 84% of respondents use or plan to use AI tools, and 46% distrust the accuracy of AI output. The top frustration, cited by 66%, is “AI solutions that are almost right, but not quite.”
If you hire that person, the Government of Canada Job Bank lists a median of $56.73 an hour for software engineers in the Ottawa region, with a low of $36.06 and a high of $88.00. The median is $52.88 in the Montreal region and $53.60 across Quebec. The figures come from the Statistics Canada Labour Force Survey for 2023 and 2024.
At the Ottawa median, a full-time software engineer earns about $110,600 a year at 37.5 hours a week, before benefits and payroll costs. That figure is our arithmetic from the hourly median. In Ontario, 96.7% of workers in this occupation receive at least one benefit on top of their wage.
No official source publishes how long a do-it-yourself AI build takes or how many hours it needs, so this page gives no estimate.
Where your data goes
For a manufacturer or distributor, the records at stake are drawings, prices, customer orders and supplier costs. Where they are processed depends on the plan and the model you choose.
With Claude Code and Anthropic
- No Canadian option on Anthropic’s API. Requests run in the US or globally, and workspace data is stored in the US only.
- Amazon Bedrock in Canada can route outside Canada. In the Canada Central region, Opus 5.5 and Sonnet 5 run only through cross-Region profiles. The US profile keeps data within US and Canada regions, and the Global profile routes requests worldwide.
- Training depends on the plan. On Pro and Max, Anthropic trains on your data, including Claude Code sessions, when the model-improvement setting is on. On Team, Enterprise and the API, it does not train on your code or prompts unless your company opts in.
- Retention depends on the plan too. On Pro and Max, Anthropic keeps your data for 5 years when model improvement is allowed, and for 30 days otherwise. Team, Enterprise and API accounts have a standard 30-day retention. Claude Code also keeps session transcripts in plain text on the developer’s computer for 30 days by default.
- Zero data retention is limited. It is available to qualified accounts on Claude for Enterprise, and Anthropic enables it after confirming eligibility. An admin cannot switch it on from settings. Fable 5 and 5.1 require data retention by default, so a zero-retention organization may not be able to use them.
If someone at your company runs Claude Code on a personal Pro or Max account with company files, the consumer training setting on that account decides what happens to those files.
With ThriveAI
The platform ThriveAI builds on is designed to keep your records on their own server in Canada that only your company uses. You choose the model that reads them: one that runs on that server, or a larger hosted model under a written agreement that the provider keeps nothing after answering. A hosted model may process requests outside Canada, so the contract names the model and its service tier.
Quebec’s privacy law applies either way
Section 17 of Quebec’s private-sector privacy act requires a privacy impact assessment before an enterprise communicates personal information outside Quebec. That includes a party outside Quebec that keeps or uses the information for you, and the act requires a written agreement that reflects the assessment. The rule applies whenever personal information leaves Quebec, whether your developer or ThriveAI builds the system.
Running it every day
A tool that your whole team uses has to run somewhere other than one developer’s computer. If your developer runs the agent on your own servers, Anthropic’s guide to hosting the Agent SDK, the kit for building agents on Claude Code, lists these jobs for your team:
- Saving each session’s state, because local disk is lost when the agent restarts.
- Monitoring, with logs sent to a collector you run.
- Sign-in for the people who use it, and the secrets and credentials it uses to reach your other systems.
- Scaling when several people use it at once.
- Tracking what it costs.
- Keeping each user’s data separate from everyone else’s.
Anthropic suggests 1 GiB of memory, 5 GiB of disk and one CPU per agent as a starting point. Anthropic also offers Managed Agents, currently in beta, where Anthropic hosts the agent loop and your application sends it requests.
Rolling Claude Code out to more than one person adds admin decisions. You choose an API provider, how settings reach each computer, which tools and commands to allow, how you see usage, and how data is handled. Some features, such as cloud sessions and Routines, need a claude.ai account and do not work with API keys alone.
With ThriveAI, every connection to your ERP only reads data, and nothing touches a live system until you approve what it will read. Someone on your team approves every draft before anything is sent or saved. Once the first tools are in daily use, ThriveAI keeps adding to them as your business changes and as AI models get better. ThriveAI holds no audited certification such as SOC 2 or ISO 27001, and each security measure is written down so your IT lead or adviser can review it.
Working with Derik to build these AI agents has been fun, very worthwhile, with a huge ROI for my team.
Fahd Alhattab, Founder, Unicorn Labs.
When to pick each
Do it yourself with Claude Code when
- A developer on your staff has time to build the system, host it and keep it running.
- Your records can be processed outside Canada, and you have done the assessment Quebec requires for any personal information.
- You want the lowest cash cost, and your developer’s time is already paid for.
- You plan to build many tools over the coming years and want that skill inside your company.
- The job is small enough for your developer to read every line and test it, such as a script or a report.
In those cases, doing it yourself is the better choice. Put the developer on a Team, Enterprise or API account, where Anthropic does not train on your code or prompts unless you opt in.
Work with ThriveAI when
- Nobody on your payroll writes software, and you do not plan to hire someone for this.
- The job draws on your ERP, email, drawings and spreadsheets together, and your team needs answers that show where each number came from.
- You want the system set up for you on a server in Canada that only your company uses, without building that hosting yourself.
- You want to see one job working, at a fixed price, before you commit to more.
- You want someone on site with the people who do the job, working in the systems you already run, including an older ERP on your own server.
To see what these systems do at each desk in a plant, read AI for manufacturing, module by module.