Questions, answered.
46 answers across the site
The basics
09 What does MagicPill Labs do?
MagicPill Labs is an AI-native software development studio in Ridgefield, Washington, founded in July 2025. It sells six service products across three stages: Strategy finds and prices the opportunity through an AI opportunity audit and fixed-price solution scoping, Build ships it as an automation or a full stack application, and Operate keeps it running under a managed agreement. AI training and consulting sits alongside those for teams who want to build the capability in-house.
How do I start working with MagicPill Labs?
Every engagement starts with a free intro call, which normally runs 30 minutes and carries no obligation. MagicPill Labs responds to enquiries within one business day. From there the usual path is a discovery audit to map the opportunity and scope the work, then the project itself; teams that already know what they want to build can go straight to scoping. There is no minimum spend required to start a conversation.
Do you bill hourly?
Yes. MagicPill Labs bills either hourly or by fixed scope, whichever fits the engagement. Fixed-scope work is quoted and agreed before it begins, and hourly work is billed against monthly timesheets so clients can see exactly where the time went.
What does it cost to work with MagicPill Labs?
Most MagicPill Labs engagements run between $1,500 and $10,000 per month. The common ones in that range are discovery audits, which map where AI and automation will pay off and scope the work, and project engagements, which build, ship, and run the automations and applications those audits identify. Where an engagement lands in the range depends on scope, pace, and how many systems are involved, and every engagement is priced before work begins.
Where does the software run once it is built?
Systems are deployed into infrastructure the client owns, and every build includes the deployment pipeline, environment setup, observability, and alerting. Hosting, patching, and monitoring can then transfer to MagicPill Labs under a managed agreement or to your own team with the documentation and runbooks that go with it. Because the system already lives in your infrastructure, there is no migration to perform if the relationship ends.
Which AI models does MagicPill Labs use?
Anthropic is the default, and MagicPill Labs switches when another model fits the job better. OpenAI, Gemini, Grok, Jev, OpenRouter, and Hugging Face are all in regular use, and open-weight models can be run on your own hardware or cloud where prompts and data must not leave your infrastructure. AI components are built model-agnostic, so the model stays a swappable part rather than a hard dependency.
Is our data used to train AI models?
No. Where MagicPill Labs transmits client data to third-party AI providers for processing, those providers are not permitted to train models on it unless that is explicitly disclosed and consented to. Client-provided data in a consulting engagement is treated as confidential, used only for the contracted purpose, and deleted on completion of the contract or on client request unless a record must be kept by law.
Do you work remotely, or only in the Portland–Vancouver area?
MagicPill Labs is headquartered in Ridgefield, Washington and works on site with clients across Clark County and the Portland–Vancouver metro, where discovery, workshops, and delivery sessions carry no travel cost. Everywhere else in the United States is served remotely. The process is identical, with discovery interviews, systems review, workshops, and delivery check-ins over video, and pricing does not change.
What happens after a system launches?
After launch the choice is a managed agreement or a full handoff. Managed operations covers continuous monitoring, scheduled health checks, incident response with a defined escalation path, dependency and security updates, scheduled improvement cycles, and a quarterly system review, priced as a retainer meaningfully cheaper than the build it follows. A handoff transfers the system with documentation and training, so your team runs it.
Solution Scoping
03 What is included in a fixed-price solution scope?
A MagicPill Labs solution scope includes a use-case analysis and feasibility review, a high-level architecture and integration outline, a full decomposition of the work into estimable units, written risk, dependency, and compliance flags, a phased roadmap, and a formal proposal ready for signature. The decomposition is the part that matters most: it is what makes the resulting quote defensible instead of a guess.
How is solution scoping different from an AI opportunity audit?
An audit looks across the whole business and asks what is worth building; scoping takes one idea and asks what it will actually take to build. The audit's output is a ranked shortlist; the scope's output is an architecture, a decomposition, and a signable proposal for a single initiative. Companies that already know what they want to build usually skip the audit and start at scoping.
What happens if the scope changes mid-build?
Scope changes are re-estimated against the same decomposition the original quote was built from, so a change is priced in the same units as the rest of the work rather than absorbed as an unpriced extra. Because work is broken into small units, most changes affect a handful of them, and you see the cost of a change before it is accepted, not on the next invoice.
Automation Solutions
03 What kinds of work can actually be automated with AI?
The work that automates well is repetitive, rule-shaped, and high-volume: intake and triage, routing, data entry and re-entry between systems, document extraction, recurring reporting and reconciliation, status chasing, and first-pass drafting or review. Work that automates badly is judgment-heavy, low-volume, or depends on context that lives only in someone's head. MagicPill Labs scopes automation candidates against that distinction before quoting anything.
What are the solution tiers, and which one will we need?
MagicPill Labs scopes automation across five tiers: tier 1 is configuration of a tool you already own, tier 2 is a reusable skill your team triggers on demand, tier 3 is that same skill running on a schedule with a human at the checkpoints, tier 4 is an agent that completes tasks asynchronously using tools and instructions, and tier 5 is a full application. Most operational problems land at tier 2 or 3, and the recommendation is always the lowest tier that actually solves the problem.
How do you handle errors and edge cases in an automation?
Every automation ships with logging, monitoring, and explicit exception handling from day one, plus human-in-the-loop checkpoints at the decisions that carry real consequences. Cases the system is not confident about are escalated to a person with the context attached rather than guessed at, and failures surface as alerts rather than as silently missing work.
Full Stack Custom Software
01 What does a typical custom software build timeline look like?
Large builds are phased, and each phase is decomposed until every unit of work is small enough to estimate honestly, so the timeline comes out of the scope rather than being set before it. A focused internal tool commonly runs a small number of weeks; a multi-surface platform runs across several phases with a working system at the end of each. Delivery is iterative, with monthly timesheets and observability included throughout.
Managed Solutions
03 What is covered under managed operations?
MagicPill Labs Managed Solutions covers continuous monitoring and scheduled health checks, incident response and triage with a defined escalation path, dependency, runtime, and security updates, a scheduled improvement cycle instead of an ad-hoc request queue, a quarterly system review with a refreshed roadmap, and optional capacity for light feature evolution.
Can you take over a system somebody else built?
Yes. Inherited systems are a standard Managed Solutions engagement, typically for software left behind by a departed contractor or a team that has moved on. Onboarding starts with a review of the live system and its architecture, after which monitoring, updates, and incident response run the same way they would on a system MagicPill Labs built.
Can we exit a managed agreement and take the system in-house?
Yes. Managed Solutions is not a lock-in mechanism, and exit means a full handoff with documentation, runbooks, and training so your team can operate the system without MagicPill Labs. Because the system is deployed into infrastructure the client owns, there is no migration to perform on exit.
AI Training & Consulting
02 Who is the AI training for — engineers or non-technical staff?
Both, in separate sessions. MagicPill Labs runs leadership briefings that give decision-makers the vocabulary and judgment to evaluate AI proposals without a translator, and practitioner workshops for the people doing the work, built on the tools and workflows they already run. The capability assessment at the start of an engagement decides which roles need which sessions.
What does the team keep after the training ends?
Teams keep the playbooks, prompt patterns, and decision frameworks produced during the sessions, written so they can be edited and extended internally rather than treated as fixed artifacts. That includes a build-buy-or-leave-it-alone framework for evaluating future AI proposals and written guidance on where a human has to stay in the loop.