A1 / APPLIED AI FOR CONSUMERS
OneShotAndy
Pflegegeld-Prüfer
HACK_002: Signal From Tomorrow
Overnight AI hackathon · Vienna · 26–27 September 2026
Thank you for being part of HACK_002.
EVENT IS OVER
ARCHIVE
{{ clock }}
TELOS CIRCLE PRESENTS
HACK_002:
SIGNAL FROM
TOMORROW
POWERED BY HOIV
Building the future we were promised.
OVERNIGHT AI HACKATHON/VIENNA/SAT 26.9. → SUN 27.9.2026
EVENT IS OVER
OPEN BRIEFING ↓
{{ countdownLabel }}{{ countdown }}
SYS.PANEL / 002{{ feedLine }}
- STATUS
- SIGNAL LOCKED
- BUILD TIME
- 19H 30M
- LOCATION
- VIENNA
- FORMAT
- IN PERSON ONLY
- LANGUAGE
- ENGLISH
- MODE
- HUMAN + MACHINE
- TEAM SIZE
- 1–4
- TRACKS
- 2 BLOCKS / 6 TRACKS
- OUTPUT
- WORKING PROTOTYPES
00
RESULTS / HACK_002
19.5 HOURS · 25 TEAMS · 25 PROJECTS SHIPPED
TRACK WINNERS
A2 / APPLIED AI FOR YOU
pJacked
Gina — cycle assistant
A3 / DEV TOOLS
8Zip
Tardis
WILDCARD
Fuck Misinformation (Science based ofc)
Foreshock
B2 / NEURAL SYSTEMS
Royal Enfly
B3 / SCIENTIFIC DISCOVERY · DOUBLE WINNERS
Kinder schoko-bons
gencite
miRacle
FRAMES / HACK_002 TRANSMISSION
01
TRACKS / TWO BLOCKS · SIX TRACKS
A1—A3 · B1—B3
Two blocks: A covers general AI and software; B focuses on science, especially biology and medicine. Each block has three tracks. The examples and resources are starting points, not requirements — teams may use other tools or datasets if they can explain access, licensing and evaluation.
BLOCK A / GENERAL AI & SOFTWARE
Useful AI Systems
A / CONSUMERSA1
A1
/ PRODUCT
Applied AI for Consumers
Build: a usable AI product for real users
Apps, websites, assistants, AI tutors, travel planners, new interfaces.
TRACK DETAILS / A1
FOCUS
Build AI-powered apps, websites, assistants, consumer services, or new interfaces that people can use today.
BUILD
- Choose a defined audience and one task they struggle to complete; specify the input and a successful outcome.
- Ship an app, website, assistant, or interface that completes the task end to end and keeps consequential choices visible to the user.
- Demonstrate realistic successes and failures; make the core flow usable by someone outside the team.
CONCRETE EXAMPLES
- A Vienna trip planner turns a budget and mobility needs into an editable route using map data.
- A language tutor detects one misconception in a spoken answer and gives a targeted follow-up exercise.
- A household assistant turns bills into draft calendar and budget entries, asking approval before saving.
STARTER RESOURCES
OpenStreetMap Overpass (map data); OpenAI API quickstart (app guide).
MEASURE
Task completion on five cases, time or quality versus a simple baseline, and feedback from at least one intended user.
A / YOU & YOUR ORGA2
A2
/ WORKFLOW
Applied AI for You
Build: a tool or workflow that improves how you or an organisation works
Internal assistants, reporting automation, research workflows, knowledge systems.
TRACK DETAILS / A2
FOCUS
Improve how you, your team, company, lab, or institution works: automate repetitive work, connect tools, manage knowledge, or support decisions.
BUILD
- Map one real workflow before and after the proposed change, including the person responsible for each decision.
- Build an internal tool or automation connected to at least one relevant data source or tool; require review before consequential actions.
- Run representative cases with consented or synthetic data and make errors, permissions, and manual overrides visible.
CONCRETE EXAMPLES
- Grant-call triage takes a call and a lab profile, then drafts an eligibility checklist with source links.
- A support inbox classifies requests, drafts a reply, and routes it to a human for approval.
- A lab inventory flow reconciles purchase emails with stock records and highlights exceptions.
STARTER RESOURCES
n8n first workflow (automation guide); OpenAI evals (evaluation guide). Use approved internal data or synthetic fixtures.
MEASURE
Time saved, error rate, and number of human corrections across at least five representative cases.
A / DEVELOPERSA3
A3
/ DEV TOOLS
Dev Tools
Build: a tool or infrastructure component for building AI systems
Codex/Jev tooling, local-model runtimes, agent infrastructure, evals, security tools, benchmarks.
TRACK DETAILS / A3
FOCUS
Build tools for developers: experiment with Jev or Codex, run local models in new environments, improve security workflows, or strengthen agent infrastructure.
BUILD
- Identify one developer bottleneck and deliver a reusable CLI, plugin, SDK component, eval harness, or runtime.
- Integrate it with a real repository or model stack and document a setup another developer can repeat.
- Compare performance, reliability, resource use, or security with a baseline and show one failure it catches.
CONCRETE EXAMPLES
- A Jev semantic lint assigns typed risk scores to pull requests and compares them with human labels.
- A Codex regression reproducer turns issue reports into failing tests and records its success rate.
- A local-model agent runtime adds resource limits, a latency dashboard, and prompt-injection checks.
STARTER RESOURCES
Jev quickstart; Codex SDK; Ollama API; OWASP GenAI Security.
MEASURE
On a fixed test repo or workload, report detection and false-positive rates, latency, cost, or resource use versus a baseline.
BLOCK B / SCIENCE, BIOLOGY & MEDICINE
AI Advances Science
B / LONGEVITYB1
B1
/ AGING
AI for Longevity & Aging
Build: a tool, model, analysis, or hypothesis related to aging and healthspan
Biomarker analysis, aging clocks, biological data, literature mining.
TRACK DETAILS / B1
FOCUS
Apply AI to aging, healthspan, biomarkers, biological data, literature, and research workflows.
BUILD
- Define one narrow, testable aging or healthspan question, including variables, population, and outcome.
- Build a reproducible notebook, model, or research tool with documented public data and a simple comparator.
- Report uncertainty and possible confounders; separate exploratory findings from clinical claims.
CONCRETE EXAMPLES
- A NHANES notebook visualizes age-stratified biomarker patterns and documents exclusions and sampling limits.
- A GEO analysis compares one aging-related expression signature across two studies and flags batch effects.
- An evidence map links proposed aging mechanisms to papers and highlights contradictory findings.
STARTER RESOURCES
CDC NHANES datasets; NCBI GEO DataSets.
MEASURE
Held-out error or an effect estimate with uncertainty; explain sampling, confounding, and why the result does not establish a treatment benefit.
B / NEUROB2
B2
/ NEURAL
Neural Systems
Build: a simulation or experiment around neural and brain-inspired systems
Neural simulation, connectomics, spiking networks, in-silico nervous systems.
TRACK DETAILS / B2
FOCUS
Explore brain-inspired computation through neural simulation, in-silico nervous systems, connectomics, and emergent behaviour.
BUILD
- Specify a neural mechanism or architecture and one measurable behaviour it should produce.
- Build a small simulation or connectome analysis with observable dynamics and a clear baseline.
- Perturb at least one parameter or connection; visualize the result and state the biological assumptions.
CONCRETE EXAMPLES
- A 100-neuron spiking network learns a simple pattern; compare it with randomized synaptic weights.
- A small connectivity graph shows how motif statistics change when selected edges are removed.
- A simulated sensory-to-motor loop plots response stability as noise and signal delay increase.
STARTER RESOURCES
Allen Brain Map (data and atlases); Brian2 tutorials (spiking-network simulator); repo: The-Uploading-Lab/fly-behavior-experiments.
MEASURE
Pattern recall, response delay, network stability, or another defined metric across a parameter sweep and baseline.
B / DISCOVERYB3
B3
/ SCIENCE
AI for Scientific Discovery
Build: a system that accelerates part of the scientific process
Scientific agents, hypothesis generation, experiment design, lab automation, automated analysis.
TRACK DETAILS / B3
FOCUS
Accelerate science with knowledge synthesis, scientific agents, lab automation, experiment design, and automated analysis.
BUILD
- Choose one stage of the research cycle and specify the artifact a scientist would inspect or use.
- Build an agent or tool that preserves source provenance, methods, and a human review step.
- Evaluate it on a fixed set of questions or records against a manual or simple baseline; show failures.
CONCRETE EXAMPLES
- An evidence graph answers one research question with traceable papers and flags unsupported claims.
- An experiment planner proposes controls and a cost-aware design for one lab question, then invites expert review.
- An automated QC notebook finds outliers in a public assay table and logs every flagged decision.
STARTER RESOURCES
OpenAlex API (scholarly graph); NCBI E-utilities (PubMed access); Open Targets API (biomedical evidence).
MEASURE
Citation support, reviewer agreement, false positives, or time saved; a generated hypothesis is not a validated discovery.
02
WHO SHOULD JOIN
ACCESS: OPEN / NO CLEARANCE REQUIRED
WHAT IS A HACKATHON?
A hackathon is a focused build sprint: small teams turn an idea into a working prototype overnight. Research, design, domain knowledge, and clear problem framing matter just as much as programming.
WHO SHOULD JOIN?
Builders, researchers, designers, founders, students, hardware people, biotech specialists, and curious people who want to make something real. Apply solo or with a team.
PRIZES
Prizes and award categories will be announced at a later date.
03
DAY 0 / REGISTER ON LUMA
SEQ.PRE-EVENT
DAY 0 WORKSHOP
The in-person Day 0 takes place on Friday, 25 September, with a talk by Irene Fialka and the AI Primer. Register on Luma; you can also join just one block. The workshop materials are public online.
IN PERSON / DAY 0FOR EVERYONE
IRENE FIALKA: TALK + AMA
Irene Fialka, former CEO of INiTS, built one of Austria's leading startup incubators and now advises top corporations. Her talk, “A problem worth solving: how science is turned into innovation”, is followed by an Ask Me Anything on startups, founding and fundraising. Open to technical and non-technical people alike. On Saturday she mentors teams at the hackathon, free of charge.
IN PERSON / DAY 0NON-TECHNICAL WELCOME
AI PRIMER
Join Peter from Machine Literacy for an AI crash course and hackathon onboarding: modern AI workflows, agentic coding, hackathon ideation, hacker culture, and where AI is heading next. Aimed mainly at non-technical participants and newcomers; everyone is welcome. Dinner break at Hofer next door.
REGISTER & PREPARELUMA / ONLINE
04
EVENT SCHEDULE / ARCHIVE
CEST / VIENNA / UTC+02:00
OFFICIAL TIMETABLE
All times are Vienna time (CEST, UTC+2). Build window: Saturday 13:00 to Sunday 08:30 — 19 hours 30 minutes.
Saturday, 26.9.DAY 01
11:00Arrival, check-in + team-finding
12:00Opening, challenges + rules (until 13:00)
13:00Building + mentoring sessions begin
18:30Dinner
19:00Overnight building continues
Sunday, 27.9.DAY 02
08:30Submission deadline
08:45Judging (until 11:00)
11:00Break + showcase preparation
12:00Winner presentations + awards
13:00Hackathon ends
SAT 13:00SUN 08:30
05
ORGANISER / TELOS CIRCLE
VIENNA / NONPROFIT / STARTUPS & RESEARCH
VIENNA / NONPROFIT / STARTUPS & RESEARCH
A society built around
difficult technical problems.
Telos Circle brings builders, researchers, and founders together around ideas worth pursuing. Events create the spark; a committed network helps turn them into projects, companies, and lasting collaborations.
06
SPONSORS & PARTNERS / SUPPORTING NODES
BECOME A SPONSOR
START A CONVERSATION →
07
PLACES / IN-PERSON & ONLINE
COORD 48.18°N 16.39°E / DISTRICT 1030
Arsenalstraße 11, 1030 Vienna. A 4,500 m² innovation hub — coworking floors, studios and creative space. Open through the night for the full build window.
POWER: YESWIFI: YESOVERNIGHT ACCESS: YES
VISIT HOIV ↗
DISCORD
The community channel for accepted participants: team-finding, updates, sleeping arrangements, and on-the-ground coordination.
ACCESS AFTER ACCEPTANCE
SUBMISSION
Upload your demo video and Git repo through our Google Form.
FOR ATTENDEES ONLY
08
FAQ / QUERY TERMINAL
Do I need a team?
No. Come alone or with people. Arrival and team-finding start Saturday at 11:00 on site. You can also find teammates earlier in Discord once you are accepted.
Do I need to be a developer?
No. Designers, product people, researchers, scientists, founders, operators, storytellers, hardware builders, and domain experts all belong here. You need curiosity, agency, and the willingness to build.
I am no hacker, why should I join?
“Hacking in a hackathon” is not about cybersecurity or “hacking into the mainframe.” It originates with hackers in the 1960s–1990s who used products in unintended ways — for example, turning computers that were originally calculators into game machines. If you have adapted a product to your needs or built a prototype, you have already been part of this hacker culture.
What can I build with?
Anything. It does not have to be software.
Do I have to attend the Day 0 workshop?
No. On Friday, Irene Fialka talks about startups and turning science into innovation, and the AI Primer presents everything you need to know to start working and hacking with AI. You can join one block or both, and register on Luma. Check out the slides at machine-literacy.com.
Where do I sleep?
Some people sleep; some do not. If you want to crash with local people, ask in the Discord after acceptance. The venue stays open throughout the night either way.
What do I submit?
Submit a short demo video and a ZIP of your Git repo through our Google Form by Sunday, 27 September at 08:30 CEST. Aim for two minutes; four minutes maximum for the video.
Do winners present at the End Show?
Winning teams present on Sunday at 12:00 CEST. We can play their submitted video, or they can show a live demo. A few questions afterwards are optional; no slides or prepared talk are needed.
Which language is spoken?
English for anything official. German happens naturally in the room.
09
JURY
MEET THE JUDGES
Judge
Judge
Judge
Judge
Judge
Judge
Judge
Judge
Judge
Judge
Judge
10
ARCHIVE / HACK_001
PREVIOUS TRANSMISSION & PHOTO LOG
