In a five-person development company, the founder may also be the architect, salesperson and project manager. A new enquiry arrives in the morning, a client needs an estimate for extra work by lunch, and the afternoon goes on explaining why the support allowance has run out. Coding waits until evening.
Small IT service providers face the same problem: user support, systems maintenance and new implementations run alongside each other. Each job may be small, but remembering every agreement and next step becomes a job of its own.
Studio OS helps organise the operations around this client work. Its business agents help process enquiries, prepare proposals and follow project progress. You set the direction, clarify the important agreements and review proposals that need a decision.
Turn an enquiry into a clear scope and proposal
“We need a self-service portal” is not yet a development task. Before promising a price, you need to know who will use it, which integrations it needs and what the first version must do. Studio's sales agent Kairi helps keep client communication and proposal preparation together. Your team confirms technical feasibility and scope.
Once the work is agreed, project agent Madis helps follow tasks, progress and blockers. Shared proposals, documents, work status and invoices can be available through the client portal, giving the customer a place to find information about their order.
Extra work deserves an agreement too. If a client requests another integration halfway through development, record its scope and the effect on price and delivery. Studio provides the client and project context; the development team estimates the change and the client accepts the proposal.
Paperclip helps manage an AI development team
Paperclip is an open-source platform for organising AI agents at work. It lets you assign roles, responsibilities, goals and tasks, and follow activity and costs. For a development company, that means a way to have several specialist agents work towards a shared outcome.
For example, a technical lead can break the work into stages, a development agent can prepare a code change, and a testing agent can check the result. An agent's ability to write code, run tests or work in a repository depends on its connected development tool and permissions. Paperclip coordinates those agents.
Tasks have owners and statuses, work can be linked to projects and goals, and agent runs can be inspected. Budgets and spending records help show what the AI work costs. Approvals give a person control over important organisational decisions.
What does Studio's Paperclip integration do today?
Studio has a Paperclip connection for agents, their runs, approvals and activity history. It can bring Paperclip agent status, budgets and spending into Studio, and let a person approve or reject Paperclip approval requests through Studio.
The current connection serves Hiiu Studio's internal development team, which builds Studio OS itself. This provides a practical starting point for designing a small development company's workflow. A customer's own Paperclip team, repository permissions and work handoffs require a separate agreement and setup.
A useful division of work would put the client relationship, proposal and commercial project status in Studio, with agreed development tasks coordinated between agents in Paperclip. The current connection does not promise an automatic journey from an accepted proposal to finished, deployed software. Code review, tests and the release decision need an explicit place in the workflow.
Start with one bounded development task whose result you can verify yourself: a bug fix, additional tests or a documentation update.
Wemply helps show where people's time goes
Using AI leaves plenty of human work to account for. Analysis, client meetings, code review, testing and support still take the team's time. If those hours go unrecorded, it is difficult to know which projects are worthwhile.
Wemply supports manual time entries and timers, including project-based tracking. It also covers work schedules, holidays, absences and payroll. For a small IT company, this provides a basis for understanding both recorded hours and team availability.
Wemply time tracking and workforce management features
Studio's HR agent Piret has a read connector for Wemply employees, schedules, absences, holiday balances, working time and wage reports. Adoption requires Wemply access and validating the queries against the company's actual data. The current connector does not create or change time entries in Wemply; automatically linking project hours to Studio invoices needs separate implementation.
An example: a small improvement to a client's portal
Imagine a four-person team whose existing client wants a CSV export in their portal. A workflow using Studio and a separately configured Paperclip development team could look like this:
- Enquiry and scope. Studio holds the client request and proposal. A developer clarifies the export fields, access rules and acceptance criteria. The client agrees to the scope and price.
- Development. The team lead gives the bounded task to a Paperclip agent. Within its tools and permissions, the agent prepares a code change and tests.
- Review. A person reviews the change, checks data visibility and confirms the tests pass. Release follows the team's agreed process.
- Time. People record analysis, development and review time in Wemply. AI spending is checked separately in Paperclip.
- Handover. The client-facing project status is updated in Studio and billing is prepared according to the agreement. The lead compares the completed work with the original estimate.
This illustrates a possible division of work. Automatic movement of tasks, hours and invoices between the three systems must be agreed separately for the company.
Look at human time and AI costs together
Paperclip agent spending and Wemply timesheets answer different questions. One records the cost of AI runs; the other records people's time. Estimating a project's delivery cost requires both, alongside other direct costs such as test environments or purchased services.
For an hourly engagement, agree with the client what you will bill. For a fixed-price job, actual time helps reveal whether the estimate's assumptions held up. The amount of AI-generated code or the duration of an agent run does not itself represent billable human hours.
One client and one recurring type of work are enough for a first pilot. Measure how long proposals take, how quickly a task reaches a verified result and whether everything needed for billing is available at month-end. Use those answers to decide what to expand next.
Studio keeps client agreements and next steps together. Paperclip helps organise AI development work. Wemply helps account for people's time. A considered division of work gives a small team more time to build solutions.