Services Consulting, retainer, automation
AI that fits the work, not the other way round.
We help teams decide where AI genuinely helps, stay alongside them while it lands, and automate the workflows that deserve it. Every engagement is scoped in writing.
01 Pricing
Three ways in. No retainer required to start.
Prices are shown in US dollars with the yuan equivalent. Sessions are invoiced after delivery; the retainer is invoiced monthly in advance.
-
HourlyUS$750/ hour¥5,000 / hour · two-hour minimum
- Working sessions with the people who own the workflow
- Written recommendation after each session
- Data boundary, review step, build-or-buy call
- Remote or on site
-
MonthlyUS$15,000/ month¥100,000 / month · month to month
- Scheduled working sessions each month
- Asynchronous review of prompts, boundaries, and outputs
- Written monthly note: what changed, what to decide next
- One named contact on each side; one team at a time
-
Per workflowScoped/ workflowFixed price after a mapping session
- Local-first or self-hosted pipelines
- Explicit review points and readable logs
- Source, boundary document, and handover
- Quoted after one consulting session
02 AI consulting
Find the right first project, and its edges.
Most AI work fails on scope, not on models. We work with your team to pick a concrete workflow, decide what a model should and should not touch, and define what "done" looks like before anyone builds.
- For
- Founders, product leads, and small to mid-sized teams that want a practical AI capability inside an existing workflow, not a platform migration.
- What you get
- A written recommendation after each session: the workflow, the data boundary (what stays local, what goes to a model, what is logged), the review step a person keeps, and a build or buy call.
- How it runs
- Working sessions with the people who do the work, remote or on site. Two hours is enough to scope one workflow; longer engagements are booked in hourly blocks.
- Pricing
- US$750 per hour (¥5,000), two-hour minimum. Invoiced after each session. Ongoing work moves to the monthly retainer; pipeline work is quoted under Workflow automation.
- To book
- Email support@kernelbeam.com with the workflow you have in mind and roughly how many people are involved.
03 AI retainer
Someone in the room while it lands.
A first project rarely fails at the kickoff. It fails in week six, when the edge cases arrive and nobody owns the decision. The retainer keeps KernelBeam alongside your team for the months that matter.
- For
- Teams that have chosen a first AI workflow and want an experienced hand on it every week until it runs without us.
- What you get
- Scheduled working sessions each month, asynchronous review of prompts, data boundaries, and outputs between sessions, and a written monthly note on what changed and what to decide next.
- How it runs
- One named contact on each side. Session cadence and the review channel are agreed at the start; scope is revisited monthly, in writing.
- Pricing
- US$15,000 per month (¥100,000), invoiced monthly in advance. No long-term commitment; either side can end at the month boundary.
04 Workflow automation
Pipelines for the work you repeat every week.
Transcription, translation, captioning, image and voice generation, formatting, publishing to several platforms: when the same sequence runs every week, it should run as a pipeline with a person checking the result, not a person doing every step.
- For
- Content teams, small studios, and operations teams with a recurring production process that is currently held together by hand.
- What we build
- Local-first or self-hosted pipelines that chain the right models and tools for each step, with explicit review points, logs you can read, and no data leaving your environment unless the step requires it.
- What you get
- The working pipeline, its source, a boundary document stating which data goes where, and a short handover so your team can run and change it.
- Pricing
- Fixed price per workflow, quoted after a consulting session that maps the current process end to end.
05 How we work
The same rules we build our apps by.
KernelBeam is a small studio. You work directly with the people doing the work, and the principles behind our software apply to our services as well.
-
I
Clear boundaries first.
Before a model sees anything, we agree on what data it may see, where it runs, and what is kept. This is written down, not assumed.
-
II
A person stays in the loop.
Every workflow we recommend keeps a human review step where it matters. AI assists; it does not sign off.
-
III
Small, then bigger.
We prefer one working workflow over a roadmap of ten. Prove it, measure it, then extend it.
06 Contact
support@kernelbeam.comTell us the workflow you have in mind and roughly how many people are involved. We reply with questions, then a session time or a scoped proposal.