Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.kowalah.com/llms.txt

Use this file to discover all available pages before exploring further.

Expert Requests are how ideas become working AI solutions. When someone on your team spots an opportunity — a process that could be automated, a workflow that could be smarter, a skill that could save hours — an Expert Request is raised and a Kowalah specialist builds it with them.
Every Expert Request lives inside a Project. Most are raised directly when the work is already scoped, but some arrive by promoting an Opportunity once it’s been triaged. An Expert Request can also optionally be linked to a specific Deliverable when it’s part of a larger piece of work.

When to raise an Expert Request

You’re managing your AI program, tracking projects and deliverables, and your team is using the platform. At some point, someone identifies something that AI could transform. That’s when you raise an Expert Request.

You know what you need, but not how to build it

Your sales lead knows exactly which part of the deal process needs a Claude skill, but they’re not the person who can build it. The Expert Request connects their domain knowledge to a specialist who can.

You have the skills, but not the capacity

Your team could build it, but they’re focused on other priorities. An Expert Request gets it built in days without pulling your people off their work.

The person closest to the problem isn't in IT

The best ideas come from the people who live the process — finance analysts, HR managers, operations leads. Expert Requests mean they don’t need to wait for IT availability or learn to build it themselves.

You want it done right, fast

Kowalah’s specialists have built hundreds of Claude skills and AI workflows. They know the patterns, the pitfalls, and the fastest path from idea to production.
Expert Requests are not just for technical builds. You can request strategic advice, training sessions, department audits, and adoption roadmaps — anything that helps your AI program move forward.

What you can request

Expert Requests cover eight categories. Here are some examples of what teams commonly ask for:

AI Agents & Skills

Build a custom Claude skill for your team, a workflow agent that automates multi-step processes, or a multi-agent system that coordinates across tools.

Training & Enablement

Run a team training session on Claude, design a multi-week training programme, or enable your Claude Guide champions with advanced skills.

Strategy & Advisory

Book a strategic AI advisory session, run a department-level AI audit to identify opportunities, or develop a full AI adoption roadmap.

Prompt Development

Create a single high-quality prompt, build a prompt collection for a workflow, or develop a full prompt library for a department.

AI Assistants

Build a knowledge assistant trained on your documents, a department-specific assistant, or an assistant with integrations to your existing tools.

Automation & Integration

Connect AI to your existing systems via API, build a data processing pipeline, or automate a multi-step business workflow.

Web Applications

Build an internal tool, an interactive calculator, or a data dashboard that brings AI-powered insights to your team.

Optimization & Support

Review the performance of existing AI solutions, upgrade a solution with new capabilities, or set up an ongoing support arrangement.
Not sure which category fits? Just describe what you need in the request — your Kowalah team will help match it to the right package.

How it works

1
Choose a service package
2
Select from categories like Claude skills and agents, prompt engineering, web applications, AI platform setup, integrations, and more. Each package describes the type of work, typical effort, and delivery timeline.
3
Describe what you need
4
Provide a title and description (minimum 20 characters). Optionally add:
5
  • Goals — what you want to achieve
  • Success metrics — how you’ll measure whether it worked
  • Timeline preference — standard or rush delivery
  • 6
    Add stakeholders
    7
    Optionally add colleagues who should be involved. For each stakeholder, you can choose whether they receive kickoff invitations, progress updates, or final deliverables.
    8
    Submit
    9
    Your request moves to In review and enters your team’s approval queue.
    10
    Approve
    11
    An admin on your team reviews the request and clicks Approve. Only then does the request enter the delivery pipeline and get assigned to a Kowalah specialist. You’re always in control of what goes into the pipeline.

    Request lifecycle

    Expert Requests move through these statuses:
    StatusWhat’s happening
    DraftYou’re still working on the request, not yet submitted
    PendingSubmitted and being triaged by your Kowalah team
    In reviewAwaiting approval from an admin on your team
    ApprovedYour team’s admin approved the request; queued for a Kowalah specialist
    In progressA specialist is actively building your solution
    Review readyWork is done, ready for you to review
    CompletedDelivered and accepted
    CancelledRequest was cancelled
    You can save a request as a draft and come back to finish it later. Drafts don’t count against your quota.

    Approving requests

    Every Expert Request has to be explicitly approved by an admin on your team before Kowalah starts work. This approval gate sits at the In review stage and means nothing enters the delivery pipeline, and no credits are committed, unless your own team has signed it off. This matters because Kowalah’s team often sits with your business users, spots opportunities, and drafts the request on their behalf. The approval gate makes sure a customer-side admin is the one saying “yes, build this”.
    1
    Open the request
    2
    Open the Expert Request detail page. If the request is In review and you have admin permissions, you’ll see an Approval panel.
    3
    Review the details
    4
    Check that the title, description, goals, success metrics, timeline, and stakeholders match what your team actually wants.
    5
    Approve or decline
    6
  • Approve — optionally add notes (for example, context for the specialist or scope clarifications). The request moves to Approved and is assigned to a Kowalah specialist.
  • Decline — provide a reason (minimum 10 characters). The request moves to Cancelled. You can always raise a new request with revised scope later.
  • Who can approve: organization admins and Core Team members. Other roles can view the request and nudge an admin to decide.
    Kowalah overrides: in rare circumstances (for example, during initial rollout or a time-critical delivery agreed by phone), a Kowalah Operations Manager or Platform Administrator can override the approval gate. Overrides are recorded on the request with the person who overrode it and their justification, so the audit trail is always visible to your team.
    Once a decision is recorded, the approval panel collapses to a read-only banner showing who approved or declined the request and when, so the governance record stays on the page for anyone who opens it later.

    Progress tracking

    Once work begins, you can track progress on each request:
    • Completion percentage — how far along the work is
    • Health status — on track, some risk, or concerned
    • Status owner — who needs to act next (Kowalah working, waiting on you, blocked externally, or in review)
    • Estimated completion date — updated as work progresses

    Quota management

    Your organization’s Expert Request quota depends on your Kowalah Managed Services plan. The dashboard shows your current usage and remaining requests.
    • Quota allocation — the number of requests included in your plan per period
    • Usage — how many you’ve submitted so far
    • Remaining — how many are still available
    • Reset date — when your quota renews (monthly, quarterly, or annually depending on your contract)
    Draft requests don’t count against your quota. The quota is only used when you submit a request.

    Stakeholders

    Add stakeholders to an Expert Request to keep the right people informed. Each stakeholder can be configured to receive:
    • Kickoff invitation — included in the initial call or briefing
    • Progress updates — notified as work progresses
    • Deliverables — receives the final output
    Stakeholder roles include primary contact, collaborator, reviewer, and approver.

    Scoring

    Every Expert Request carries an ICE score — Impact, Confidence, and Ease, each on a 1 to 10 scale. Two scores are tracked:
    • Estimated score — set when the request is being scoped, captures what the team expected to deliver
    • Final score — set at closeout, captures how the work actually landed
    The gap between the two is a useful signal: where final Impact came in lower than estimated, the team’s scoping can sharpen next time; where it came in higher, the request unlocked more value than anyone predicted. For the full scoring model, see ICE Scoring.

    Outcomes

    When an Expert Request is delivered, the realised business results — money saved, hours freed, contracts won — are captured as Outcomes on the request. Outcomes can be added at closeout or any time afterwards as new evidence materialises, and customer-side stakeholders linked to the request can capture them too.

    Connections to other features

    Expert Requests are connected to:
    • Project — every Expert Request lives inside a Project
    • Deliverable — optionally linked to a specific Deliverable when the request is part of a larger piece of work
    • Source Opportunity — if promoted from an Opportunity, the request carries a back-link to it
    • Tasks — requests can have sub-tasks for tracking individual work items
    • Outcomes — realised results captured at closeout and beyond

    Deliverables

    Committed work inside a project, often with linked Expert Requests

    ICE Scoring

    How estimated and final scores work across the platform

    Outcomes

    Capture the realised results of a delivered request

    Tasks

    Track work items within a request