Software Development · San Francisco, CA
AI and Software Engineering for San Francisco Teams
For a San Francisco team building with AI, the hard part is rarely calling a model. It is knowing whether the feature is right often enough to ship, keeping it right after the next model release, and finding engineers to do the work around it. That is where an outside team can help.
Schedule a call for your San Francisco project →What can an outside firm add to a San Francisco engineering team?
Capacity and focus. An outside firm can take a scoped piece of the roadmap, such as an MVP, a mobile app or a stalled codebase, or it can do the evaluation and guardrail work that moves an AI prototype into production. The test is whether your engineers can own the result once the engagement ends.
Four ways we work with teams that already have engineers
- →Engineers inside your team. They work in your repository, your tickets and your review process. See staff augmentation.
- →A codebase that has stalled. We look at what is there, then finish it or stabilize it. See software project takeover.
- →A first version of a new product. Scoped to what real users need to try, built to grow. See MVP development and SaaS development.
- →An AI feature with a deadline. The eval and guardrail work below, attached to your product.
Getting an AI feature from demo to production
A demo proves the model can do the task once. Production needs proof that it does the task often enough, on your data, and keeps doing it.
- →Evaluation sets from your own material. Every AI build starts with a set of real examples from your domain, with agreed correct answers, and the set is re-run on every change. A prompt edit, a retrieval change or a model upgrade gets a score before it ships.
- →Guardrails on agents. Tools are allowlisted, any irreversible action waits for a person to approve it, and every tool call is logged so a bad outcome can be traced.
- →Provider choice per engagement. We build with Anthropic's Claude and OpenAI models, called directly or through your own cloud when your data has to stay there. See our Anthropic and OpenAI pages.
- →Model costs after launch. We do not charge for model usage while we build; once the AI is part of your product, your company pays the provider for its usage directly.
California's rules for automated decisions
The California Privacy Protection Agency's updated CCPA regulations took effect on January 1, 2026. They give consumers rights to access information about, and to opt out of, a business's use of automated decisionmaking technology, and they require certain businesses to complete risk assessments and annual cybersecurity audits. Businesses that use ADMT to make significant decisions must meet the ADMT requirements from January 1, 2027.
If your product helps decide who gets credit, housing, a job, an education opportunity or health care, the opt-out path, the explanation a consumer can request and the record of what the model saw are product features. On regulated builds we scope them in the design phase, with your counsel deciding which obligations apply.
Working across three time zones
San Francisco is three hours behind Mt Pleasant, South Carolina, where we are based. Our late afternoon is your lunchtime.
Services we offer in San Francisco
AI Implementation
Custom AI systems built around your existing workflows. Average 30–60% reduction in manual process costs.
Learn more →AI Workflow Automation
Automation pipelines built around the systems you already run, eliminating manual data entry and cross-system copying.
Learn more →Legacy Modernization
Replace outdated software with modern, maintainable code.
Learn more →Custom Software Development
Full-stack web apps, APIs, and integrations built to your spec, from MVP to enterprise-grade production systems.
Learn more →How an engagement works
01
Workflow Audit
We map your existing processes against current AI capabilities in the 90-minute AI workflow audit, identifying exactly where automation creates measurable ROI. The audit is complimentary when it is bundled with a larger build and available as a paid one-off on its own.
02
Build Sprint
A scoped build against a written scope. Prototype, staging, and deployment milestones are agreed in the written scope. Most large builds are fixed-price, with a monthly retainer after launch.
03
Launch + Support
Deployment and handover. Post-launch support and maintenance are scoped per engagement.
Common questions from San Francisco businesses
Do you charge for model usage on AI projects?
Not while we build: model usage during development is part of the build. Once the AI feature is live in your product, your company pays the model provider for that usage.
Can you take over a codebase another team started?
Yes. We look at what is there before anything is priced, then agree a written scope to finish or stabilize it.
How do you know an AI feature is good enough to ship?
We build an evaluation set from your own examples, with agreed correct answers, and score the feature against it. The same set is re-run on every change, so a model upgrade or prompt edit is measured before release.
Do California's automated decisionmaking rules apply to our product?
They apply to businesses covered by the CCPA that use automated decisionmaking technology for significant decisions, with compliance required from January 1, 2027. Your counsel decides whether your product is covered; we design the opt-out, access and record-keeping pieces it would need.
Ready to start your San Francisco project?
Book the 90-minute AI workflow audit. The audit is complimentary when it is bundled with a larger build and available as a paid one-off on its own.
Schedule a call for your San Francisco project →