Media resources · Facts and references
Unfire.
Public facts, clearly stated.
Unfire builds independent software and creative technology, from controlled AI workflows and alternative computer interaction to audio tools and digital experiences.
Last reviewed: 21 November 2026. This page distinguishes shipped products, experimental research and third-party listings. It does not imply endorsement by any platform or institution.
Media enquiries by emailFast facts
Unfire
Independent, founder-led software and creative technology initiative based in Catalonia, Spain. Not separately incorporated as a company at this stage.
Omar Baró
Product development, research and creative software. Read the founder profile.
Bolt & Unfire Control
Experimental software for human-supervised AI workflows and exploratory hands/voice-based computing interaction.
Written-first communication
Products and research status
- Bolt: controlled AI workflow prototypes featuring explicit permission boundaries, human review and audit records. Early-stage/internal demonstrations are not independently certified or enterprise production deployments. Project overview.
- Unfire Control: research-stage exploration of hands and voice interaction, including potential accessibility applications. It is not clinically validated, a certified assistive device or a publicly approved healthcare solution. Research overview.
- Archer: published audio software, including a Mac App Store listing and a separate iPhone/iPad listing. Availability varies by region.
- Power Energy Prices and Unfire Pocket Capture: publicly listed energy information and audio recording tools.
- Games and interactive experiences: see the official overview and itch.io profile; check each product page for availability.
Independently verifiable references
European Commission I3 portal
A published partner-search profile describes Unfire's proposed controlled AI workflow contribution. See the published record.
APTE / PTE DISRUPTIVE
Unfire's technology collaboration offer was confirmed as published by APTE on 19 November 2026. Visit the public marketplace.
Devpost · BOLT PostCode
An externally accessible technical project showcase. Read the project page.
GitHub repositories
Public source and documentation for selected research demonstrations. View the public engineering profile.
A third-party listing is not evidence of an awarded contract, membership, certification, endorsement or independent performance evaluation.
Independent award recognition
BOLT Trusted AI Bridge — Best AI Startup, Bronze
The organizer lists BOLT Trusted AI Bridge by Omar Baró / Unfire as a 2026 bronze winner. Verify the official winner record. The recognition is based on submitted project materials, not independent software-security certification, a signed customer contract, public funding, government endorsement or incorporation of Unfire as a company.
New: inspectable AI research evidence
Research update · 21 November 2026 · Public and reproducible within the stated limits
The BOLT Agent Control Plane is a distinct, MIT-licensed, read-only developer experiment. It takes structured AI action proposals and applies deterministic policy rules that classify them as allowed, requiring review or denied. It does not execute external actions, operate a live approval service, establish commercial production readiness or certify AI safety.
v0.1.0 pre-release
Browse the dated, source-linked research release, not a production software launch.
12 offline checks passed
Local automated checks were reported as 12/12 passing on the specified public commit. Review the technical evidence brief for the test scope and limitations.
Software Heritage snapshot
Inspect the preserved public-source snapshot. Archiving preserves source history; it is not an audit, endorsement or a certification.
Separate from commercial products
This read-only experiment is not the shipping Bolt macOS application, Unfire Autonomy Convergence or Unfire Control Vision. No confidential product code is included in the public materials linked here.
For a concise technical explainer, read Who authorises an AI agent’s next action? — a source-linked research note with the public experiment’s scope, limitations and practical evaluation questions.
Potential editorial angles
These are suggested topics for editorial investigation, not claims of completed independent validation.
- What human approval boundaries and recovery controls can add to AI-assisted software workflows.
- The practical challenges of camera and voice interaction in accessibility-oriented computing research.
- How an independent developer approaches building and releasing audio, utility and interactive software across different platforms.
Press contact and verification
For factual checks, technical background, screenshots or an asynchronous interview, write to hello@unfire.technology. Communication is handled by email; no live call is required.
For additional public evidence, check the source-linked public record, research methodology, and relevant product pages. Editorial teams should verify all current claims and product availability directly with the linked sources.