Nothing leaves the phone.
Capture, fit assessment, tailoring, and the screener loop run locally through Apple Foundation Models or the deterministic extractive engine. The current app has no network entitlement.
The iPhone app turns a job description into a tailored résumé and interview brief, every claim grounded in your real experience. The upcoming 1.4 update adds local profile import and ATS-safe export.
Cloud résumé tools upload your work history to someone else's servers and hand back confident prose you can't fully vouch for. SiloApply takes the opposite stance: it runs entirely on the device in your pocket, and it refuses to write a claim it can't trace back to something you actually did. Tailored to the role, grounded in your record, kept on your hardware.
Capture, fit assessment, tailoring, and the screener loop run locally through Apple Foundation Models or the deterministic extractive engine. The current app has no network entitlement.
Each tailored bullet links to a real entry in your experience, and a deterministic validator confirms it. Fabricated numbers are rejected before they reach the page. Defensible in the room, not just on paper.
No sign-up, no subscription, no profile on a server you don't control. The $24.99 one-time unlock works on iPhone today and will carry to the Mac app when it launches.
Snap or paste a job description and SiloApply rewrites your résumé for it, but every bullet carries a citation back to a real entry in your experience. A deterministic validator confirms the grounding and rejects any fabricated metric before you ever see it. Tailored, never invented.
From the same JD and your profile, SiloApply drafts likely questions, grounded talking points that point back at your actual work, the gaps worth pre-empting, and notes on company fit. A prep sheet that knows what you can defend.
The upcoming 1.4 update reads PDF, scanned PDF, image, RTF, and plain-text résumés locally. It shows the exact extracted source spans for your review before anything becomes citable evidence. DOCX is not imported.
Version 1.4 adds a text-based, single-column PDF and equivalent Markdown, plus a separate preparation file. Standard headings, no tables or graphics, and concrete readiness checks replace made-up ATS percentages.
SiloApply ranks your real accomplishments against the job, then asks the on-device model to write bullets that reference only that evidence. Generation happens per section, sized to the model's context, so nothing is invented to fill space.

SiloApply role-plays the screener the role implies (a frontier-lab recruiter reads differently than a defense program lead) and returns an honest verdict with the screen-killer gaps. Then it revises and screens again, until the draft would pass.

The AI critique is advisory. What is authoritative is a deterministic pass that verifies every bullet's citation, rejects any unsupported number, and flags weak grounding for your review, plus a linter that strips marketing fluff and overclaims. Truth is enforced by code, not hoped for.

Cut p95 cold-launch latency 38% by rewriting the on-device cache layer.
Shipped 4 App Store releases owning the on-device Vision OCR pipeline.
Mentored engineers across the platform team on Swift concurrency.
Version 1.4 accepts PDF, scanned PDF, image, RTF, and plain text. Extraction and OCR happen on-device; DOCX remains intentionally unsupported.
Imported facts must resolve to exact spans in the original text. You select, correct, and confirm them before they can become citable evidence; generated paraphrases are never imported as facts.
Required, preferred, logistics, compensation, responsibility, and culture signals are separated into a requirement-to-evidence view. Hard gaps stay distinct from optional gaps before tailoring begins.
The 1.4 export uses a text-readable, single-column PDF with standard headings and an equivalent résumé Markdown file. Preparation and evidence stay in a separate Markdown file.
Readiness reports concrete results: PDF text extraction, reading order, page count, standard sections, keyword coverage, linter findings, and provenance state. There is no universal ATS percentage.
SiloApply role-plays the role's likely screener, critiques the draft, and revises. The AI critique remains advisory; deterministic citation and language checks decide what is verified.
Apple Foundation Models runs on supported devices. Otherwise SiloApply uses its deterministic extractive engine. Both paths stay on-device and pass the same provenance checks.
Everything lives encrypted on your device. Move a profile between iPhone and Mac with an encrypted export over AirDrop. No server, no account, ever.
Profile import, role screening, readiness preview, and the first tailoring run stay free in version 1.4. The existing $24.99 one-time unlock adds ATS PDF export and unlimited local tailoring. It applies to iPhone now and the Mac app when that release arrives.
View the iPhone app ↗SiloApply is available on iPhone today. Version 1.4 is the next iPhone update, adding local résumé import, fit-first role screening, and ATS-safe PDF and Markdown export. Mac remains coming soon until its separate App Store review clears.